tax changes in very different economies
TRANSCRIPT
Georgia State University Georgia State University
ScholarWorks @ Georgia State University ScholarWorks @ Georgia State University
Economics Dissertations
Summer 2014
Tax Changes In Very Different Economies Tax Changes In Very Different Economies
Jeffrey Condon
Follow this and additional works at: https://scholarworks.gsu.edu/econ_diss
Recommended Citation Recommended Citation Condon, Jeffrey, "Tax Changes In Very Different Economies." Dissertation, Georgia State University, 2014. doi: https://doi.org/10.57709/5778454
This Dissertation is brought to you for free and open access by ScholarWorks @ Georgia State University. It has been accepted for inclusion in Economics Dissertations by an authorized administrator of ScholarWorks @ Georgia State University. For more information, please contact [email protected].
ABSTRACT
TAX CHANGES IN VERY DIFFERENT ECONOMIES
By
JEFFREY THOMAS CONDON
August 2014
Committee Chair: Dr. Andrew Feltenstein Major Department: Economics
Despite the prevalence of computable general equilibrium (CGE) models applied to tax
changes of varying types, little work has been done focusing on state level comprehensive tax
reform or on tax reform in countries undergoing a regime change. This research develops and
applies methodologies for analyzing fiscal policy changes under these two very different
economic scenarios. The findings for each application are relevant to policy makers as they
weigh the effects of tax reform. The models developed for the two scenarios offer guidance to
future modelers in studying similar economies and the contrast of the two provides a
framework for thinking about model design and application. Finally, the results, when
compared to each other, allow us to see the relative effectiveness of the two tax reform
policies given their very different economies.
A method for extending Lofgren’s 2002 standard CGE model for developing countries is
developed to analyze state personal and corporate income taxes, sales taxes, and taxes on
intermediate goods. This extension is then used to study comprehensive state level tax reform
in Georgia. The model is run in GAMS using 2010 IMPLAN data with industries aggregated
specifically for the tax reforms. The tax reforms analyzed include and shift from the income tax
to the sales tax as considered by the Georgia legislature in 2011 as well as an extension of that
policy which totally replaces the income tax with a sales tax. Results are reported for short and
long run effects on gross state product, consumer income and equivalent variation, industrial
output, and the state’s budget.
A model containing tax evasion and tax avoidance behavior is developed and applied to
Egypt in which the economic effects of the regime change are simulated within the duration of
the dynamic model. The model is run using 2008 data from the Egyptian Central Bank and
includes an exogenous shift in foreign demand for Egyptian exports to simulate the effects of
the regime change. Results are presented for the growth paths and end states for consumer
income and welfare, gross domestic product, inflation, and tax evasion.
TAX CHANGES IN VERY DIFFERENT ECONOMIES
BY
JEFFREY THOMAS CONDON
A Dissertation Submitted in Partial Fulfillment of the Requirements for the Degree
of Doctor of Philosophy
in the Andrew Young School of Policy Studies
of Georgia State University
Copyright by Jeffrey Thomas Condon
2014
ACCEPTANCE This dissertation was prepared under the direction of the candidate’s Dissertation Committee. It has been approved and accepted by all members of that committee, and it has been accepted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Economics in the Andrew Young School of Policy Studies of Georgia State University.
Dissertation Chair: Dr. Andrew Feltenstein Committee: Dr. Sally Wallace
Dr. Jorge Martinez-Vazquez Dr. Florenz Plassmann
Electronic Version Approved By: Dr. Mary Beth Walker, Dean Andrew Young School of Policy Studies Georgia State University August 2014
iv
DEDICATION This dissertation is dedicated to my wife who deserves this dedication more than anyone, my family who have been supportive my entire life, and my friends who threatened to call me ‘Doctor’ whether I finished this or not. It is also dedicated to everyone who, while watching a race, yells “Looking good…. Keep it up…. You can do it!” rather than “Is everything ok? Why are you taking so long? When are you going to finish?”
v
ACKNOWLEDGMENTS
I would like to thank my dissertation committee members and readers, Sally Wallace, Jorge Martinez-Vazquez, Florenz Plassmann, Mark Rider, Dave Sjoquist, and Felix Rioja for their advice and patience. I would like to thank Melissa Rockwood for her hard work proof reading. I would like to thank Nour Abdul-Razzak for her early work on Egyptian tax evasion. I would like to thank my late colleague and friend Manal Metwaly for her hard work gathering data in the middle of a revolution and for being such a gracious host. I would like to thank Georgia State’s International Center for Public Policy, the director Jorge Martinez-Vazquez, and the entire staff for generously providing funds and logistical support for trips to Kenya and Egypt. I would like to thank the Andrew Young School for funding my studies. I would like to thank all of the professors who have taught me something during these fascinating years. Finally, I would like to thank my advisor Andrew Feltenstein for encouraging me when I struggled, pointing me in the right direction when I got confused, always having an open door, and for showing me the pyramids.
vi
TABLE OF CONTENTS
CHAPTER ONE: COMPARISON OF METHODOLOGIES ............................................................. 1
I. INTRODUCTION ................................................................................................................... 1
II. CONTRASTING MODEL DESIGNS ......................................................................................... 3
A. Fiscal and Monetary Policy Control ................................................................................ 5
B. Trade ............................................................................................................................... 6
C. Economic Stability .......................................................................................................... 8
D. Closure Rules ................................................................................................................ 10
III. EMPIRICAL LITERATURE .................................................................................................... 13
A. State Level Literature ................................................................................................... 13
B. Event Literature ............................................................................................................ 16
CHAPTER TWO: THE EFFECTS OF TAX REFORM IN GEORGIA.................................................. 19
I. INTRODUCTION ................................................................................................................. 19
II. GEORGIA’S TAX STRUCTURE AND REFORMS .................................................................... 20
III. SOCIAL ACCOUNTING MATRIX .......................................................................................... 21
IV. GENERAL EQUILIBRIUM SPECIFICATION ........................................................................... 23
A. Consumption ................................................................................................................ 24
B. Production .................................................................................................................... 26
C. Taxes ............................................................................................................................. 31
D Trade ............................................................................................................................. 36
V. CALIBRATION..................................................................................................................... 37
vii
A. Tax Revenue ................................................................................................................. 37
B. Parameters ................................................................................................................... 38
VI. SPECIAL COUNCIL SIMULATIONS ...................................................................................... 39
VII. SPECIAL COUNCIL SIMULATION RESULTS ......................................................................... 43
A. State Results ................................................................................................................. 43
B. Household Results ........................................................................................................ 45
C. Industry Results ............................................................................................................ 47
VIII. INCOME TAX ELIMINATION SIMULATIONS ....................................................................... 50
IX. INCOME TAX ELIMINATION RESULTS ................................................................................ 53
A. State Results ................................................................................................................. 53
B. Household Results ........................................................................................................ 55
C. Industry Results ............................................................................................................ 59
X. CONCLUSIONS ................................................................................................................... 62
CHAPTER THREE: A SIMULATED FUTURE OF EGYPT ............................................................ 65
I. INTRODUCTION ................................................................................................................. 65
II. EGYPT BEFORE AND AFTER THE REVOLUTION ................................................................. 66
III. EGYPTIAN TAX EVASION AND INFORMAL SECTOR ........................................................... 71
A. Tax Evasion ................................................................................................................... 71
B. Egyptian Informal Sector .............................................................................................. 74
IV. GENERAL EQUILIBRIUM SPECIFICATION ........................................................................... 76
A. Dynamics ...................................................................................................................... 77
B. Production .................................................................................................................... 77
viii
C. Banking ......................................................................................................................... 82
D. Consumption ................................................................................................................ 85
E. Government ................................................................................................................. 86
F. The Foreign Sector ........................................................................................................ 88
V. DATA SOURCES ................................................................................................................. 89
A. Production .................................................................................................................... 90
B. Effective Tax Rates ....................................................................................................... 91
C. Consumption ................................................................................................................ 91
D. Initial stocks .................................................................................................................. 91
E. Estimation of Behavioral Equations ............................................................................. 92
VI. SIMULATIONS .................................................................................................................... 94
VII. RESULTS............................................................................................................................. 94
A. Gross Domestic Product ............................................................................................... 95
B. Inflation ........................................................................................................................ 96
C. Budget .......................................................................................................................... 97
D. Consumers .................................................................................................................... 99
E. Formal Sector ............................................................................................................. 101
VIII. CONCLUSIONS ................................................................................................................. 103
CHAPTER FOUR: SUMMARY AND CONCLUSIONS .............................................................. 105
APPENDIX .............................................................................................................. 108
APPENDIX A – INCOME TAX CALIBRATION CHECK ................................................................. 108
APPENDIX B - SPECIAL COUNCIL SIMULATION GAMS CODE .................................................. 109
ix
APPENDIX C - INCOME TAX ELIMINATION GAMS CODE ......................................................... 109
APPENDIX D – GEORGIA SAM ................................................................................................. 111
APPENDIX E – EGYPTIAN SAM................................................................................................. 111
BIBLIOGRAPHY ......................................................................................................... 112
VITA……...................................................................................................................119
x
LIST OF TABLES
TABLE 1 – AGGREGATED INDUSTRY LABELING KEY ................................................................................... 26
TABLE 2 – DATA EXTRACT AND BUDGET COMPARISON .............................................................................. 37
TABLE 3 – PARAMETER VALUES ........................................................................................................... 38
TABLE 4 – HOUSEHOLD TAX RATES AND INCOME PER CAPITA ..................................................................... 41
TABLE 5 – BASELINE AND SIMULATED STATE GDP ................................................................................... 44
TABLE 6 – BASELINE AND SIMULATION TAX REVENUE BY SOURCE ................................................................ 45
TABLE 7 – BASELINE AND SIMULATION UTILITY BY HOUSEHOLD .................................................................. 46
TABLE 8 – EQUIVALENT VARIATION BY HOUSEHOLD SOURCE ...................................................................... 46
TABLE 9 – AGGREGATED INDUSTRY LABELING KEY ................................................................................... 47
TABLE 10 – OUTPUT BY COMMODITY ................................................................................................... 49
TABLE 11 – OUTPUT BY ACTIVITY ......................................................................................................... 49
TABLE 12 – SALES TAX REVENUE BY TREATMENT ..................................................................................... 51
TABLE 13 – BASELINE AND SIMULATED STATE GDP ................................................................................ 53
TABLE 14 – BASELINE AND SIMULATION TAX REVENUE BY SOURCE .............................................................. 54
TABLE 15 – BASELINE AND SIMULATION UTILITY BY HOUSEHOLD ................................................................ 56
TABLE 16 – EQUIVALENT VARIATION BY HOUSEHOLD ............................................................................... 57
TABLE 17 – AGGREGATED INDUSTRY LABELING KEY ................................................................................. 59
TABLE 18 – OUTPUT BY COMMODITY ................................................................................................... 60
TABLE 19 – OUTPUT BY ACTIVITY ......................................................................................................... 61
TABLE 20 – AVERAGE PRIVATE SECTOR GROWTH RATE OF COMMODITIES AND ACTIVITIES ............................... 62
xi
TABLE 21 – EGYPTIAN ECONOMY BEFORE AND AFTER ‘05 TAX REFORMS AND ‘11 REVOLUTION ....................... 70
TABLE 22 – REAL INCOME IN PERIOD 8 BY CONSUMER ........................................................................... 100
TABLE 23 – UTILITY IN PERIOD 8 BY CONSUMER ................................................................................... 101
TABLE A1 – CHECK FOR ACCURACY OF INCOME TAX CALIBRATION............................................................. 108
xii
LIST OF EQUATIONS AND FIGURES
EQ 1 – HOUSEHOLD CONSUMPTION OF MARKET COMMODITIES ............................................................... 25
EQ 2 – VALUED ADDED ...................................................................................................................... 27
EQ 3 – INTERMEDIATE INPUTS ............................................................................................................. 27
EQ 4 – CES FUNCTION ...................................................................................................................... 28
EQ 5 – MARKETABLE COMMODITIES PRODUCED BY ACTIVITY .................................................................... 29
EQ 6 – DOMESTIC COMMODITY OUTPUT AGGREGATION FUNCTION ........................................................... 29
EQ 7 – FIRST-ORDER CONDITION OF DOMESTIC COMMODITY AGGREGATION FUNCTION ................................ 30
EQ 8 – HOUSEHOLD DIRECT TAX RATE .................................................................................................. 32
EQ 9 – TAX RATE ON CAPITAL ............................................................................................................. 33
EQ 10 – HOUSEHOLD CONSUMPTION OF MARKET COMMODITIES ............................................................. 34
EQ 11 – INTERMEDIATE USE OF COMMODITY BY ACTIVITY ........................................................................ 35
EQ 12 – TRADE CALCULATION FOR CURRENT ACCOUNT BALANCE .............................................................. 36
EQ 13 – HOUSEHOLD DIRECT TAX RATE ................................................................................................ 40
EQ 14 – VALUE ADDED ...................................................................................................................... 78
EQ 15 – COST OF BORROWING AND PRESENT VALUE OF RETURN TO NEW CAPITAL EQUALITY ........................ 79
EQ 16 – INTEREST RATE ..................................................................................................................... 80
EQ 17 – PRESENT VALUE OF RETURN TO NEW CAPITAL INEQUALITY ........................................................... 80
EQ 18 – PRESENT VALUE OF RETURN TO NEW CAPITAL INEQUALITY ........................................................... 80
EQ 19 – TAX RATE OF THE FIRM ......................................................................................................... 81
EQ 20 – REAL VALUE OF CAPITAL STOCK ............................................................................................... 82
EQ 21 – DEMONSTRABLE VALUE OF CAPITAL ......................................................................................... 83
xiii
EQ 22 – CAPITAL REQUIREMENTS AND DEMONSTRABLE VALUE OF CAPITAL INEQUALITY ................................ 84
EQ 23 – BANK WILLINGNESS TO LEND .................................................................................................. 84
EQ 24 – CENTRAL GOVERNMENT DEFICIT .............................................................................................. 87
EQ 25 – DEFICIT COMPOSITION ........................................................................................................... 87
EQ 26 – DEFICIT FINANCING ............................................................................................................... 88
EQ 27 – EXPORT EQUATION ............................................................................................................... 88
EQ 28 – LONG RUN DEMAND FOR REAL MONEY .................................................................................... 93
FIGURE 1 – REAL GDP GROWTH RATES ............................................................................................... 95
FIGURE 2 – NORMALIZED REAL GDP GROWTH RATE ............................................................................. 96
FIGURE 3 –ANNUAL INFLATION RATES ................................................................................................. 97
FIGURE 4 – TAX REVENUE TO GDP RATIO ........................................................................................... 98
FIGURE 5 – DEFICIT AS A PERCENT OF GDP .......................................................................................... 99
FIGURE 6 – TAX COMPLIANCE RELATIVE TO THE BASELINE: REVOLUTION .................................................. 102
FIGURE 7 – TAX COMPLIANCE RELATIVE TO THE BASELINE: REVOLUTION WITH TAX CUT .............................. 103
1
CHAPTER ONE: COMPARISON OF METHODOLOGIES
I. INTRODUCTION
Since the advent of Computable General Equilibrium (CGE), the models have been
modified and adapted to study numerous alternative policy treatments. In addition to the
expanded capabilities of the models themselves, there has been a dramatic increase in the
amount of resources given to the collection of data required for these models. With
increasingly detailed datasets and new and interesting applications created each year, there are
many opportunities for previously unexplored policy analyses. This dissertation considers the
use of CGE models to analyze fiscal policy changes in two regions with very different economies.
Focusing on fiscal policy, we have a wide spectrum of regional or policy applications
available to us. Taxation and government expenditure analyses using CGE have been conducted
across the globe for years. In each case, the model’s general design must be modified according
to the level of regional fiscal control, monetary control, and trading partners, calibrated to the
state of the economy, and must take into account any exogenous factors that are essential to
the model. In this research, we look at two extreme ends of the spectrum: the state of Georgia
within the United States of American (USA) and the nation of Egypt.
These two economies present an interesting contrast of CGE application as they are so
drastically different in a few interesting ways. Georgia operates within the USA’s currency while
2
Egypt, as a sovereign nation, controls its monetary policy. The USA--and as a result Georgia--has
free trade within the nation and federal control of its policies regarding trade are above
Georgia’s level of control, while Egypt has unilateral control of its trade policy. Georgia’s tax
structure represents a relatively modest portion of the total fiscal picture, as federal and local
taxes are not under its control, while Egypt is a single fiscal zone. Georgia’s economy is stable
within the business cycle, while Egypt has suffered the consequences of a political revolution
and regime change with dramatic impacts on its trade, currency, and largest industries.
The purpose of this research is to explore the application of CGE methodology in fiscal
policy analysis as applied to the subtle changes resulting from modest revenue neutral changes
of a currency within a region and a free trade zone, and then compare that approach to a
model that assumes a large exogenous change to a country. This chapter will look at the
contrasting specifications required in the Georgia and Egypt models. Because we are focusing
on fiscal policy in a CGE framework, this chapter will begin by comparing and contrasting the
complexities of CGE modeling in these two different economies. Chapter Two of this
dissertation will present a subnational model of tax reform for the state of Georgia. Chapter
Three will present a national model for Egypt, evaluating alternative fiscal policies to address
the exogenous effects of the revolution. Chapter Four contains conclusions.
3
II. CONTRASTING MODEL DESIGNS
Chapters Two and Three detail each model’s specifications and the methodology used
for evaluating the treatments. They also present background information about the status of
the economies and supporting empirical literature for similar models. Before presenting the
methodology in its entirety we address the two models generally, to note important differences
in what is required in each case, and provide some useful background.
The Georgia analysis is a static simulation of a tax reform package for which the defining
characteristic is a decrease in the income tax, with a corresponding expansion of the sales tax
to previously untaxed sectors, thus maintaining revenue neutrality. In Georgia, the purpose of
the analysis is to predict the effects of the policy on welfare and output. In Egypt, the analysis
is sequentially dynamic, incorporating an informal sector, with an exogenous shift of foreign
demand to simulate the political revolution, and the treatment is a hypothetical fiscal policy
response. The purpose of the Egyptian analysis is not to predict the future state of the Egyptian
economy or of the Egyptian people, but rather to compare the relative effects of alternative
policies in the extreme conditions of the post-revolution economy.
In any model, the designer must consider the relevant policy levers available to policy
makers, the important economic indicators used for policy evaluation, as well as the direct
effects of any exogenous changes in the model. Regarding policy levers in Georgia, we first
consider the initial tax structure and proposed reforms. The initial tax structure varies from the
reformed tax structure only with regard to the rates and their application in untaxed sectors, so
this implementation is limited only by the availability of industry and household data.
4
Furthermore, Georgia does not have control of its trade or monetary policy, so those policy
levers are not of interest in this instance. Regarding Egypt, the nation has complete control of
its fiscal, monetary, and trade policies, as long as its trading partners and lenders are modeled
realistically so as to avoid corner solutions. In Egypt’s case, we will have to pay close attention
to each decision, allowing the model to simulate the widest array of fiscal policy changes and
monetary shifts, as well as consequences of the revolution.
In Georgia, we are interested in the economic indicators affected by tax reform. Thus,
we maintain endogeneity for variables of interest such as household utility, industry output,
gross domestic output of the state, and prices. To keep to a square system of equations, some
conditions must be imposed on any model. In Georgia, for example, we maintain a balanced
budget so that government expenditures are allocated according to the previously observed
shares of the budget and exactly equal to the tax revenue generated by the model. In Egypt,
we are interested in macroeconomic indicators for purposes of policy evaluation, but we also
consider which variables are exogenously shifted to simulate the revolution. In this case we
look at indicators such as the budget deficit, currency flows, GDP, household utility, and prices.
We will look at endogenous macroeconomic variables after we exogenously adjust taxes. We
also can maintain control of variables hit hardest by the revolution, including foreign demand
for Egyptian exports such as tourism, or foreign direct investment, or the interest rate on
government debt.
5
A. Fiscal and Monetary Policy Control
CGE models are powerful tools for analyzing policy changes under the weakest ceteris
paribus conditions. However, when designing the model, we must isolate the policy
instruments available to policy makers according to their level of control. Generally, we can
separate these into three different classifications: control of both fiscal and monetary policy;
control only of fiscal policy; and control only of monetary policy. All countries with their own
currency fall into the first category where they control both fiscal and monetary policy.
Subnational regions that wield fiscal authority have control of their regional fiscal policy, with
some constraints, but do not have control of their monetary policy. Examples include states or
cities within the United States, or countries within the Economic and Monetary Union of Europe,
or countries that have dollarized, such as Panama. The final of the three categories applies only
to the European Monetary Union, as it has control of monetary policy for its region, but no
direct control of the fiscal policy of its members. The structure of the CGE model varies greatly
depending on the level of fiscal and monetary control.
For regions with control of their monetary policy, the model’s design must align with the
currency policy of the region. For countries with a floating exchange rate, we allow relative
values of currency to adjust endogenously, so that we can monitor the effects on the currency
in the simulation. For countries with a fixed exchange rate, we can hold constant the relative
values of currency, and see the effects on reserves in the central banks. In the case of Egypt,
where the exchange rate floats within a narrow range, we have chosen to maintain a fixed
exchange rate so that we can control it both before and after the regime change. Had the
6
exchange rate encountered drastic swings during the revolution, adjustment of this variable
would have been possible. Regardless, this approach allows us to allocate foreign aid to
simulate the Egyptian Central Bank’s defense of the currency.
Within the USA, and in the state of Georgia, we maintain the same currency for all
goods by allowing borrowing and savings between states to adjust and account for terms of
trade between the states. This procedure essentially gives us a fixed exchange rate between
Georgia and the rest of the USA. A similar procedure is done regarding the borrowing and
savings from the foreign sector as the effect of the revenue neutral tax change in Georgia has a
negligible effect on the value of the US dollar.
Regarding tax changes, Egypt is the simpler of the two as it is a single fiscal zone.
Changes within the Egyptian model are simply introduced for the country and applied using
constraints. Georgia, however, requires separation of the federal and local taxes in the social
accounting matrix (SAM), so that state level tax changes are not applied to federal or local tax
codes. This is a straightforward procedure that is considered when the structure of the SAM is
established.
B. Trade
Computable general equilibrium (CGE) models are frequently used for analysis of
unilateral trade reforms, multilateral tariff and quota reforms, and regional free trade
agreements. CGE models are capable of identifying the subsets of each regional economy that
7
will be helped or hurt by trade reform. As with production technology and consumption
preferences, trade data is retrieved from the social accounting matrix (SAM). Trade specific
models, depending on the structure, require trading partner specific data on import and export
tariffs and quotas, freight-on-board prices, transportation costs, and economic distance. When
regional SAM data is denominated in different currencies, nominal exchange rates are required.
This nominal rate is used to calculate the real exchange rate, which is the ratio of the prices of
traded goods to non-traded goods. The nominal rate can be flexible, fixed, or stochastic,
depending upon the context. In models that incorporate trade but for which trade is not the
primary point of interest, the nominal exchange rate is flexible and allowed to fluctuate to
permit for a fixed current account balance. This closure rule, an exogenous current account and
endogenous nominal exchange rate, applies to most countries, but this too can be adjusted
depending upon the context.
An additional factor to consider when looking at trade is the relative importance of the
trading regions. If a region is of insufficient size to impact global prices, modelers usually allow
world prices to remain fixed regardless of changes in the small country. However, if trade is
modeled using an Armington import function, each country is a monopoly provider of its goods.
If Armington elasticities are set incorrectly, unrealistically large swings in prices and exchange
rates can occur regardless of the size of the country. Furthermore, if the substitutability of
goods is parameterized incorrectly, these price swings can generate unrealistic production or
consumption decisions.
In our Georgia model, we allow the state of Georgia to be the monopoly provider of
differentiated goods and services to its trading partners: the rest of the United States (RUS) and
8
the rest of the world (ROW). For both the RUS and the ROW, we maintain a stable currency and
allow foreign savings and investment and the current account to fluctuate. This is appropriate
in Georgia, as the fluctuations have a negligible effect. Its largest trading partner is the RUS,
which uses the dollar, and that dollar is nearly unaltered by the new flow of goods and services
in response to the tax change in Georgia.
In Egypt, we model a single trading partner—the ROW—in a simple export equation in
which aggregate demand for exports is determined by domestic and foreign price indices as
well as world income. Thus, the combination of the export equation and domestic supply
responses determines aggregate exports. Demand for imports is endogenous and is derived
from the domestic consumers' maximization problems. Foreign lending is assumed to be
exogenous. Thus, gross capital inflows are exogenous, but the overall change in reserves is
endogenous with a fixed exchange rate.
C. Economic Stability
We say the economy is stable when there are no fundamental changes in the nature of
the economy during the time period covered by the calibration and simulations. Descriptions
of the Georgian and Egyptian economies as well as the calibration and simulation of the models
are explained in detail in Chapters Two and Three. For now, suffice it to say that the
fundamental structure of the economy as represented by Georgia’s SAM is unchanged for the
duration of the simulation, while Egypt’s SAM is fundamentally altered within the model’s
duration by the regime change.
9
In the case of the stable Georgia economy, we are interested in the effects of the tax
change, so that change in Georgia is the cause of the effects we are interested in studying. Thus,
we calibrate the model of Georgia to the pre-change equilibrium, institute the change, and then
look at the differences. This is the most common application of CGE models.
In Egypt, the revolution is a known confounding change we must account for in the
calibration. In this situation, we calibrate our model using a pre-revolution SAM so that it
simulates the pre-revolution economy. Without an exogenous intervention, the model would
simply simulate the economy as it would have existed in the absence of the revolution. To avoid
this, we use a sequentially dynamic model calibrated to the economy as it existed when the
SAM was established, and then adjust exogenous variables to simulate the macroeconomic
effects of the regime change. This calibration is considerably more difficult than the case of
Georgia, as we are calibrating the model, first to the economy that spawned the SAM, but then
again to a similar yet new post-revolution economy. Finally, with a fully calibrated model, we
then simulate treatments for comparison.
Another alternative is available and is used by modelers who do not know the stability
of the economy with certainty. A modeler may allow for stochastic changes to parameters.
This application of CGE models, first used by Kydland and Prescott, is called dynamic stochastic
general equilibrium (DSGE) modeling (1982). In a DSGE, the modeler addresses the Lucas
critique by allowing some parameters such as technology, prices, preferences, or government
policies to adjust randomly (1976). This approach can incorporate more sophisticated
assumptions about the rationality of institutions. For example, a consumer’s consumption and
savings decisions in the present depend upon his/her rational expectations about the real
10
interest rate, and that real interest rate is heavily affected by the stochastically determined
future nominal interest rate. In this example, we see the relationship between the expectations
of the consumer, the volatility of interest rates, and economic performance. This inclusion of
expectations has led many central banks to use DSGE models to study monetary policy. Another
feature of this approach is that it can incorporate stickiness in parameter adjustments. For
example, one might use a DSGE model to study the different effects of volatile inflation on
industries whose prices vary with respect to their flexibility or stickiness.
D. Closure Rules
Closure rules refer to the modeler’s decisions about which variables are endogenous and which
variables are exogenous. This is necessary to maintain a square system in which the number of
unknowns does not exceed the number of equations thus allowing for mathematical
techniques to solve the system of equations. Closure rules are generally divided into two
categories: micro and macro. Micro closure rules refer to the mechanisms by which individual
markets adjust to reach equilibrium. Sen (1963) notes that the system of equations is over-
determined if we impose the following: fixed investment, consumption is determined by
income; prices equal their marginal product; and factor supply equals factor demand. To avoid
the over-determined system, we must make some assumptions about the mechanisms. To
illustrate micro closure rules, it is useful to look at a single market. In the labor market, for
example, two common closure rules are neoclassical and Keynesian. Under a neoclassical
11
closure rule, labor supply is the exogenous variable, while the endogenous wage variable
adjusts until supply and demand equal. Under the Keynesian closure rule, the (sticky) wage is
the exogenous variable, and the endogenous labor supply variable adjusts until supply and
demand equal. In this example, it is easy to see that the choice of closure rule can have a
significant effect on the results of the model. There are other configurations of labor market
closure rules such as Kaldorian, which allows the wage rate to deviate from the marginal
product of labor, and Johansen, which adjusts residual consumption to obtain labor market
equilibrium. One should choose the closure rule based on the economy and the market
structures being modeled because each of these may alter the resulting equilibrium depending
upon regional and sectoral factor mobility, the relationship between savings and investment, or
other factor market traits (Dewatripont and Michel, 1987).
In addition to the micro closure rules of the factor markets, a modeler must choose
macro closure rules. Macro closure rules refer to the savings-investment balance, the
government budget balance, and the trade balance. In each case a modeler must assign
exogeneity or endogeneity. For example, investments are either endogenous and savings
driven or savings can accommodate an exogenously fixed level of investments. Models
frequently have balanced budget conditions on government so that expenditure equals
revenue, but an alternative to that approach allows for the government to access capital
markets to cover a deficit. Yet another way to approach the government balance is to allow
some tax instrument to adjust endogenously to maintain an exogenous government budget.
Regarding trade closure in a multiregional model, the trading partners each have their own
SAM, so their behavior is modeled with their own micro closure rules. In a model with a single
12
region we use that region’s SAM with import and export flows to complete the model. In either
case, one must accommodate for changes in the relative value of currencies and foreign savings
such that one adjusts to clear the current account of the balance of payments. This is frequently
done with the ROW reacting to the regions with SAMs to borrow or loan, consume or provide,
and appreciate or depreciate as needed to close the model. There is no ‘correct’ closure rule: a
structuralist may use Keynesian rules, while a neoclassical economist may use a price-
adjusting closure rule. The choice is up to the modeler’s judgment as to what most accurately
represents the economy (see Robinson and Lofgren, 2005).
In Georgia we vary the closure rules to simulate the effects of the policy change
depending upon the treatment of factors. This approach allows us to use a static model to draw
conclusions about the long run and the short run. We vary labor market and capital market
closures by adjusting the supply of either factor and the specificity of capital. In the variation
simulating the shortest run, we set capital to be activity specific with a fixed supply, and set a
fixed supply of labor. In the short run simulation, the supply of both factors remains fixed, but
capital is mobile. In the long run simulation, both capital and labor are mobile and the supply of
each is variable. Regarding macro closures, we let the government expenditure equal
government revenue with an exogenous tax structure, investment adjusts to accommodate
savings, and both the RUS and ROW savings adjust to accommodate a fixed exchange rate. The
first macro closure is crucial as it allows us to control the tax rates. The remaining two rules
have little impact on the simulation results, as the effect of the policy change on these variables
is minimal.
13
In the Egyptian model, capital is sector specific while labor is mobile between sectors
and thus the supply is variable. The supply of sector capital in period two is determined by
sector savings in period one. This calculation of the endogenous savings variable is explained in
Chapter Three. Egypt’s government may run a deficit which is financed through monetary
expansion or borrowing. The ROW is modeled as a single trading partner whose consumption of
Egyptian goods adjusts to accommodate domestic supply and whose supply of foreign goods to
Egypt adjusts to accommodate domestic consumption. The balance of payments is kept
through adjustments to the endogenous central bank reserves.
III. EMPIRICAL LITERATURE
A. State Level Literature
For the first time, both state level data and software capable of computing the
numerical solutions of CGE models were widely available in the late 1980s. IMPLAN, a private
data provider, began publishing I-O matrices in 1988, and General Algebraic Modeling Software
(GAMS) made programming and solving these problems possible outside of the base
programming languages such as Fortran or COBAL. This led to early models of Oklahoma (Koh,
1991), Southern California (Robinson, Subramanian, and Geoghegan, 1993), and Ohio (Kraybill
and Pai, 1995). Koh’s model looked at the Oklahoma’s ‘boom and bust’ cycle from 1977 to
1986. Robinson, Subramanian, and Geoghegan looked at air pollution in the LA basin. Kraybill
14
and Pai developed the model to look at infrastructure, but their two-factor model served as a
platform for follow-up papers on tax incentives and public goods (see Seung and Kraybill, 1999,
2001). Another early model is the Oklahoma model by Vargas, Eliecer, and Schreiner (1999)
which treats the factor market for wood as a monopsony. This extensively documented model
has been used in numerous follow-up papers on such topics as oligopsonies and sport fishing.
In 2002, Lofgren, Harris, and Robinson, through the International Food Policy Research
Institute (IFPRI), published a standard model along with a detailed model description and its
accompanying code in GAMS. Lofgren also developed a series of CGE exercises in GAMS to
familiarize new modelers with GAMS. As a result of Lofgren’s extensive documentation and
GAMS coding exercises, numerous researchers use this as a base. The IFPRI model has been
modified and used for hundreds of CGE applications. For example, the IFPRI model was adapted
for regional modeling by Holland, Stodick, and Devadoss and applied to the economic effects of
Mad Cow Disease in Washington (2004). In this iteration, the code modifications and additional
details regarding the new model design were published for others to build on. This process of
modifying the existing model for other purposes and publishing the documentation continues
today (See Watson et al., 2012). The Lofgren model, while the most popular, is not unique.
Similar detailed documentation and code is available for more than 40 different models within
GAMS alone. The wide distribution of so many resources has made CGE models significantly
more accessible for researchers. The result is that their use has spread, and today it is used to
model almost anything that can be modeled, from fisheries in Alaska (Seung and Waters, 2010)
to the economic impacts of forest bioenergy in Florida (Huang et al., 2012).
15
Despite their increased use for various academic purposes, academics represent a small
portion of the state analysis that takes place when proposed policies are being evaluated.
Berck, Golan, and Smith (1996) built the California Dynamic Revenue Analysis Model (DRAM) to
fulfill the state government’s requirement that all tax changes be dynamically scored. The
model ran three hypothetical scenarios, found relatively modest deviations from static methods,
was defunded in 2000 without having simulated any real policies, and is now used by
California’s Air Resource Board to study the environment (Vasche, 2006). Massachusetts also
developed a sophisticated model in cooperation with REMI, a private company. REMI operates
it today and has numerous states and cities as clients. IMPLAN, the provider of the I-O matrices
in every state-level paper mentioned above, does policy evaluation for every level of regional
aggregation down to the single county. These private firms conduct a significant share of the
modeling for which there is funding available.
Despite their small market share, academics continue to apply CGE models to study
state tax issues. Morgan, Mutti, and Partridge (1989) studied the long-run impact of local, state,
and federal taxes on factor allocation. They found that unilateral removal of distorting taxes
led to higher growth in the region and inflows of capital and labor. Waters, Holland, and Weber
(1997)1 examined the impact of Oregon’s “Measure 5,” a property tax limit passed in 1990.
They find that output and income increased after the limitation was passed, and high income
households benefited more than low income households while state and local government
1 To illustrate the repeated modifications to existing models: the Waters, Holland, and Weber model was built
using some early GAMS code by Krybill and Pai. The Julia-Wise, Cooke, and Holland model was built using the Waters, Holland, and Weber model’s treatment of the property tax. The Faulk, Thaiprasert, and Hicks model was built using the Holland, Stodick, and Devadoss model.
16
expenditures and revenues decreased substantially. Julia-Wise, Cooke, and Holland also looked
at property taxes by modeling a 50% cut to property taxes in Idaho (2002). Their study found
that $2 of tax revenue would be lost for every $3 of property tax reduction, with a significantly
regressive distribution of the benefits. Edmiston (2002) used a CGE model to simulate the
strategic apportionment of state corporate income taxes. She finds that the first mover has
significant budgetary gains by strategic apportionment, but after neighboring states respond,
the effects are diminished. However, the economic development gains remain. Dakhlia and
Strauss (2003) looked at sales and use taxes in e-commerce between Washington and Oregon.
They find the classic beggar-thy-neighbor results. Faulk, Thaiprasert, and Hicks (2013) simulated
a cap on Indiana’s property tax and a corresponding revenue increase in sales taxes. In their
study, they find long run increases in employment, income, and investment.2
B. Event Literature
There is a large body of literature using CGE models to simulate the effects of
exogenous events. For our purposes, we exclude events with a stochastic component. In 1997,
the News South Wales treasury partnered with the Center for Regional Economic Analysis to
look at the state and national economies for the 6 years prior to and the 6 years after the 2000
2 No additional scientific literature was found on CGE models studying state level tax reforms. However, there are
numerous policy evaluations done by REMI, IMPLAN, TRAIN (Tax and Revenue Analysis in Nebraska), STAMP (State Tax Analysis Modeling Program) out of Suffolk University, and others. With each of these organizations, the model description is to some degree publically available, but the applications are not peer reviewed.
17
Sydney Olympic Games (Madden and Crowe). They predicted that the significant gains would
go, almost exclusively, to New South Wales, the Australian state where Sydney is located.
Similar studies have been conducted for every Olympics going back Los Angeles in 1984.
Kasimati (2003) evaluates ex-ante studies for every summer Olympics from 1984 until 2000 and
kindly notes their predictions were ‘optimistic’. Porter and Fletcher (2008) look back and
evaluated models simulating the 1996 summer and the 2002 winter Olympic Games in Atlanta
and Salt Lake City. They concluded that because the I-O models used constant factor prices and
technical coefficients, they overstated the effects. Madden (2006), who had previously
employed I-O models for this purpose, responded to the criticism, “I would contend, however,
that the problems with I-O analysis are of such a fundamental nature that it should no longer be
used for mega events of the scale of the Olympic Games… However, I-O is not the only method
available for undertaking economic impact analysis. A more modern and much more
sophisticated method, namely computable general equilibrium (CGE) modeling, is available.”
CGE has largely replaced I-O analysis as a result of the I-O models’ public failures. For a recent
meta-analysis of impact studies on a variety of major sporting events see Li and Jago (2012).
There are also many examples of negative events being modeled. Pambudi, McCaughey,
and Smyth (2009) used the Indonesian version of TERM, called EMERALD, to construct a static
model of Indonesia’s 26 provinces. In this paper, Pambudi et al. simulated the effects of the
2002 bombing in Bali that killed 202 people. The authors exogenously “shift the export demand
schedule for tourism commodities uniformly by 50% to the left.” This exogenous shift to the
tourism industry, of which Bali was the largest producer, had effects throughout the provinces
and on the country’s terms of trade and exchange rates.
18
Looking at Australia and using a Monash University model, Horridge et al. (2005) used
TERM to study the 44 regional effects of the 2002-03 drought, which was the worst drought in
two decades. They find that effects on some statistical divisions are extreme, with income
losses of up to 20%. Despite the relatively small share of agriculture in Australian GDP, the
drought reduces GDP by 1.6%, and contributes to a decline in unemployment and to a
worsening of the balance of trade.”
Kawai and Zhai (2009) built a custom model in GAMS with 17 regions focusing on the
Asian block. Their model simulated integration and economic interactions of the export driven
Asian economies during the “evaporation” of US consumption. They find that “a combination of
domestic reform aimed at boosting service sector productivity and external liberalization aimed
at fostering broader economic integration” will be critical as the Asian economies rebalance
after an end to the export-driven growth period of the past few decades.
Additional topics studied using this methodology in CGE include the impact of El Niño on
Mexican agriculture (Harris and Robinson, 2001); the effect of remittance shocks on El Salvador
(Morley, Piñeiro, and Robinson, 2011); the effect of changes in the international prices of
tobacco and petroleum products on Malawi (Löfgren et al., 2001); the economic impact of AIDS
in Cameroon (Kambou, Devarajan, and Over, 1992); and the effect of coups in Fiji (Narayan and
Prasad, 2007). In addition to these economic impact studies of real exogenous events,
numerous papers used this method for estimating resilience to disasters such as an earthquake,
terrorist attack, or water disruption (Rose and Liao, 2005). For an overview, refer to the work
of Dwyer, Forsyth, and Spurr titled, “Assessing the Economic Impacts of Events: A Computable
General Equilibrium Approach” (2006).
19
CHAPTER TWO: THE EFFECTS OF TAX REFORM IN GEORGIA
I. INTRODUCTION
Georgia tax reform is a fresh topic. Georgia’s Special Council on Tax Reform and Fairness
for Georgians (Special Council) issued its final report on January 7, 2011. This final report
outlined a revenue neutral tax reform policy with specific tax rate adjustments, exemptions,
and inclusions. These changes followed the basic premise that the tax structure should be
broad based with low rates. They recommended expanded application of the sales tax to
personal and household services and household food consumption; a reduced personal income
tax rate with less deductions; a lower corporate income tax rate; the elimination of sales tax
exemptions on inputs for the education, health, and movie industries; and a new exemption on
energy used in the agriculture, mining, and manufacturing industries. These approaches for
reforming the tax code in Georgia were in keeping with the Special Council’s stated ‘guiding
principles’ and represented a shift away from the income tax and towards consumption-based
or use taxes.
This research will simulate the Special Council’s approach using a subnational CGE
model of Georgia. The simulations will include an expansion of sales tax to previously untaxed
sectors, proposed tax exemptions, proposed eliminations of exclusions, and the reduction of
the personal and corporate income taxes. In addition to the simulation of the Special Council’s
proposals, this paper will also simulate the elimination of the personal and corporate income
20
taxes, which are replaced with an expanded sales tax so that the tax change is revenue neutral.
The effects of each of these tax policy changes are evaluated using total economic output of the
state, state tax revenue, industry output, and household utility.
II. GEORGIA’S TAX STRUCTURE AND REFORMS
Georgia’s tax structure, as is the case in many states, is complicated. Georgia collects
revenue from a combination of personal and corporate income taxes, sales tax, and taxes on
motor fuel, tobacco, alcohol, and insurance. Property taxes are used at the local level. While
combining a number of taxes for revenue collection is common, the implementation of the tax
code in Georgia is confounded by exemptions, deductions, rate schedules, and tax holidays.
The personal income tax (PIT) in Georgia has a top marginal rate of 6% since the 1955
elimination of the 7% rate. With the exception of that change, the income schedule and
associated tax rates have remained unchanged since 1937. Income tax is levied on incomes as
low as $750 at a rate of 1%, and the top marginal rate of 6% is reached for a single filer at
$7000. Standard deductions in Georgia are significantly lower than the federal deductions, at
$2700 for filers and $3000 for dependents. There are also 28 additions or subtractions
necessary to get the Georgia adjusted gross income (AGI) from the federal AGI. The PIT
accounts for almost half of the total tax revenue for the state government. The corporate
income tax (CIT) has changed significantly over the years. There have been numerous
deductions and credits implemented to stimulate various sectors. The current rate is 6%. The
21
CIT raises less than 5% of the state’s total tax revenue. The sales and use taxes in Georgia raises
approximately 33% of total state revenue. The state sales tax rate is 4%, although with local
option sales taxes used throughout the state, the rate can reach as high as 8%. Georgia also
offers a sales tax holiday lasting three days which applies to school items, clothing, and energy
or water efficient products.
The Special Council proposed reforms to address each of the aforementioned taxes as
well as a number of others including taxes on cigarettes, insurance premiums, E-commerce, etc.
They did so under their own established ‘guiding principles’ that the tax structure should be
growth enhancing, efficient, stable, clear, fair and equitable, properly developed, and have an
avenue for resolution of disputes. Their final report was considerably more detailed than is
summarized here, but from a broader perspective it constituted a shift from the income tax
toward the sales and use tax as the primary vehicle of revenue generation for the state. They
recommended a lowering of the PIT ultimately to 4% which would be applied as a flat rate. The
CIT would also fall along with the PIT to 4%. To make up for the lost revenue, they also
recommended an expansion of sales and use taxes on services, food, and inputs for health,
education, and movies.
III. SOCIAL ACCOUNTING MATRIX
A social accounting matrix is a framework of economic data for a geographic region in
which each cell represents a payment from the account of its column to the account of its row
resulting in a square matrix of data in which expenditures are in the column and incomes in the
22
row.3 Within the SAM framework, because of double entry accounting, the sum of an account
row equals the sum of its account column. Columns and rows are similarly labeled according to
whatever the desired disaggregation of activities, commodities, factors, institutions, and both
foreign and domestic trade.
The SAM for this paper was extracted from IMPLAN using 2010 data and aggregated in
GAMS4. The extracted data originally included 440 activities and 440 commodities. These were
aggregated to 15 of each. The three factors include labor, capital, and indirect taxes. There are
16 institutions: nine households organized by income; three federal government institutions for
non-defense, defense, and investment; three state government institutions for non-education,
education, and investment; and one aggregation of enterprises, investment, and inventory
deletions. The last two rows and columns include foreign trade and domestic trade. The final
aggregated SAM is 51 rows x 51 columns. Individuals and firms are aggregated according to
their income group or industry, respectively, and these aggregations are modeled as if the
entire income group was a single representative consumer or the entire industry was a single
representative firm. It is important to note that a commodity’s production is not limited to that
corresponding industrial activity; for example, the mining activity can produce a manufacturing
commodity. Aggregation is necessary so that the solution algorithm is capable of finding a
convergent solution.
3 When referencing a specific cell in the SAM, we will refer to it using matrix notation so SAM(i,j) will refer to the
cell in the SAM in the row i and column j.
4 See Appendix D for information regarding the Georgia SAM.
23
IV. GENERAL EQUILIBRIUM SPECIFICATION
The model used here is based on the International Food Policy Research Institute (IFPRI)
CGE model described in Lofgren et al. (2002). It was later adapted for regional modeling
purposes by Holland, Stodick, and Devadoss in their Washington-Idaho model (2004) to include
domestic and foreign trade. This model was developed in General Algebraic Modeling System
(GAMS) and thus adaptable for alternative purposes. Because the CGE model is a system of
linear and nonlinear equations in which the number of equations is equal to the number of
endogenous variables in the model, the PATH solver, which is a generalized variation of
Newton’s method, is used to calculate the solution.
When accounting for each individual commodity, activity, factor, household,
government unit, and foreign sector, the model contains 839 individual equations and 839
individual variables. These equations and variables are divided into four blocks: a production
and trade block, an institution block, a price block, and a constraints block. The
production and trade block refers to the equations that reveal the endogenous equilibrium
quantities. The institution block governs the behavior of the institutions which include
households, government, and enterprises and investment. The price block refers to the system
of equations that reveals the endogenous equilibrium prices. The constraint block ensures a
square system of equations by enforcing constraints such as the factor demand equals the
factor supply.
This model does not contain an objective function. Its equations define the behavior of
factors in the economy such that their behavior is subject to first order optimality conditions.
24
The model assumes that producers and consumers will maximize profit and utility, given prices,
endowments, and technology. The maximization is subject to constraints for the behavior of
markets and governmental units as well as macroeconomic flows such as savings, investment,
or foreign demand. Households provide the factors—labor and capital—to firms that use these
factors as well as intermediate inputs to produce the final output. Intermediate goods are an
Armington composite of domestic and imported goods.
Trading partners include the RUS and foreign trade. The final output generated is
allocated to domestic consumption and exports. Households consume final goods and public
services, and either save or borrow the remainder of final output. The government consumes
intermediate goods, produces public services, and distributes money in the form of transfers to
institutions. The government tax revenue includes all transfers from institutions and all revenue
generated by taxes on intermediate and final goods. In any CGE model, there are more
equations than are possible to detail in an applied paper. However, because this paper looks at
the effects of tax reform on households, firms, and the government, it is important to consider
the utility-maximizing households, profit-maximizing producers, and the treatment of taxes and
government within the model.
A. Consumption
Consumers are divided into nine households according to income level. Households
receive income from factor endowments and from transfers from other households,
government, and enterprises. Household income is used to pay taxes, to fund transfers to
25
other households, to consume domestic market commodities and foreign goods, and to save.
Household income available for consumption is the income received from factors and transfers
net of taxes, transfers to other institutions, and savings. Household consumption of market
commodities is governed by:
( ∑ ( )
) ( ( ))
(1)
where,
QHc,h = the quantity consumed of commodity c by household h,
ϒc,h = the subsistence level of consumption of market commodity c for household h,
βc,h = the marginal share of consumption spent on market commodity c for household h,
EHh = net income of taxes, savings, and transfers,
PCc = the composite price of commodity c,
tqc = the sales tax on commodity c.
This is a linear expenditure system function and the first-order condition for a Stone-
Geary utility function, which governs the utility-maximizing consumers. Multiplying both sides
of the equation by the relevant prices would derive the total household spending on marketed
commodities. Note that activities are not included in this equation as there is no home
consumption at the cost of production in the SAM. Also, the prices are those paid by
consumers for the composite good of domestic and imported commodities.
26
B. Production
The SAM is aggregated to 15 industries for both activities and commodities. There are
two factors: labor and capital. A producer is represented as an activity and is modeled to
maximize the difference between revenue earned and the cost of factors of production and
intermediate inputs. Revenue is earned by selling the commodities produced by an activity. This
profit maximization is subject to the technology. All participants in the market are price takers.
Table 1 lists the 15 industrial commodity and activity aggregations.
TABLE 1 – Aggregated industry labeling key
Aggregation Label Aggregation Description
AFFH Agriculture, forestry, fishing, and hunting MIN Mining
MANUF Manufacturing MEDIMAN Medical equipment manufacturing
UTIL Utilities FOOD Groceries for home consumption
INFOSER Information services HEEDSER Health and education services OTHSER All other services CONST Construction
WHOLTRAD Wholesale trade RETTRAD Retail trade
TRAN Transportation MOVIE Motion picture and music video production
GOVSPC Government services
The technology is modeled by a nested function with constant elasticity of substitution
(CES) and Leontief functional forms contained within. To produce output, the firms use factors
and intermediate inputs. Using factors, we derive the value added using a CES function.
Intermediate input of commodity c used by activity a is derived using a Leontief function. The
27
value added and intermediate inputs are then combined in the top level of the nested function
using a CES functional form for the final activity level. Each representative firm chooses its
optimal activity level by optimizing the combination of value added and intermediate inputs
such that they are consumed until their marginal revenue product is equal to their price. This
process reveals the activity level of the firms.
Valued added is derived in Equation 2:
(∑
)
(2)
where,
QVAa = quantity of aggregate value added by activity a,
α = value added shift parameter,
δ = value added share parameter,
QFf,c = quantity demanded of factor f for activity a,
ρ = transformation of the elasticity of factor substitution.
Here we see that the value added is derived from a CES function of disaggregated factor
quantities.
Intermediate inputs are derived using Equation 3:
(3)
where,
QINTa,c = intermediate demand for commodity c from activity a,
28
φ = quantity of commodity c used as an intermediate good per unit of activity a,
QINTAa = quantity of aggregate intermediate input for activity a.
Here we see that firms’ demand for specific commodities as intermediate inputs for their
activity is derived using a Leontief function with a fixed intermediate input coefficient.
Next in the production function we must combine the value added and intermediate
inputs using a CES function:
(
( ) )
(4)
where,
QAa = activity level for activity a,
α = activity shift parameter,
δ = activity share parameter,
QVAa = quantity of aggregate value added for activity a,
QINTa,c = intermediate demand for commodity c from activity a,
ρ = is a transformation of the value added and inputs elasticity of substitution for
activity a .
Equation 4 is the CES function at the ‘top’ of the nested function. Below this are the CES value
added function and the Leontief intermediate inputs function, Equations 2 and 3, respectively.
The optimal choices of value added and intermediate inputs, and thus factors and aggregate
intermediate inputs, is derived by their relative prices. This is the primary mechanism by which
industrial output is altered by policy simulations.
29
With all factors combined with all inputs to produce an activity level, we can determine
the quantity of marketable commodities produced by the activity:
(5)
Where,
QXa,c = marketed quantity of commodity c from activity a,
Øa,c = fixed yield coefficient for the quantity of commodity output c per unit of activity a,
QAa,c = production of commodity c from activity a.
This configuration allows each activity to produce any commodity and each commodity to be
produced by any activity.
Equation 6 is the domestic commodity output aggregation function:
(∑
)
(6)
where,
QXc = aggregated marketed quantity of domestic output of commodity c,
α = shift parameter,
δ = share parameter,
QXa,c = marketed quantity of commodity c from activity a,
ρa,c = domestic commodity aggregation function exponent of activity a for commodity c.
In Equation 6 we see that the aggregate marketed quantity of a commodity is a CES aggregate
of the marketed output of the activities producing the commodity.
30
The first-order condition of the domestic commodity aggregation function is:
(∑
)
(7)
where,
PPa,c = producer price of commodity c for activity a,
PPc = aggregate producer price of commodity c produced from all activities,
QXc = aggregated marketed quantity of domestic output of commodity c,
δ = share parameter,
QXa,c = marketed quantity of commodity c from activity a,
ρa,c = is a transformation of the elasticity of substitution of activity a for commodity c.
In Equation 7 the marginal cost of commodity c from activity a equals the marginal revenue
product of commodity c from activity a.
Together Equations 6 and 7 are the first-order conditions for our profit-maximizing
representative firms selling their optimized activity at the market price PC subject to the
aggregation function and the disaggregated commodity producer price PP. Regarding the prices
in the consumption and production equations, note that PCc, the composite price of commodity
c; PPc, the aggregate producer price from all activities for commodity c; and PPa,c, the producer
price of commodity c for activity a, can be different from one another and those differences
may change as the model allows for trade and tariffs, as well as taxes according to activity or
31
commodity. Demonstrated here are three of the 10 prices included in the model. Relationships
between the 10 different modeled prices are contained in the 10 equations of the price block.
C. Taxes
Government revenue in a generic SAM format and the model can be derived from
numerous sources: direct taxes on institutions (households and enterprises), direct taxes on
factors, value added taxes, activity taxes, import tariffs, export taxes, sales taxes, factor income,
and transfers from other governmental units and the ROW5. This shows the versatility of the
format and the Lofgren model. However, when looking at Georgia, not all of these apply.
Georgia’s revenue is derived using direct taxes on individuals and enterprises, taxes on factors,
indirect taxes, and transfers from the federal government. For our purposes, federal
government transfers are left unchanged and we focus on the taxes altered by the proposed
reform: income tax, sales tax, and taxes on intermediate inputs.
In the model we simulate a change in the income tax by adjusting the ratio of transfers
to the state as a proportion of total expenditure. However, total expenditure equals total
income, and household income is derived from the labor endowment as well as the capital
endowment, transfer payments, and dividends. An additional complication arises because the
5 For brevity we omit the equation for total government revenue, but for each source of tax revenue the tax rate is
either applied to the stock, as in the case of factors, or applied to the product of the price and the total for that source, as in the case of a value added tax (Revenue = Tax * (Price of Value Added * Total Value Added). These taxes can be institution, factor, activity, commodity, or trading partner specific.
32
direct transfers from households to the state include not only income tax, but also property tax,
estate tax, and some minor fees such as hunting licenses. These calculations are explained in
the simulations section. Regarding the model, the simulation is done by declaring a variable:
( ))
( ) (8)
where,
tyh = direct tax rate given by the SAM for a household h,
SAM (SG, H) = transfers from households H to the state government SG,
SAM (Total, H) = total expenditure for a household H.
The value of this parameter is then controlled in the counterfactual simulation. We cannot
simply declare a variable for the ratio of transfers from households as a proportion of labor
income (though we tried) because the direct tax rate on households is used repeatedly
elsewhere in the model. How this is used to adjust the tax on the factor labor is explained in the
simulations section.
The corporate income tax (CIT) is similarly problematic as it encompasses transfers from
the capital factor to the state and transfers from enterprises to the state. Compounding the
problem is the aggregation of enterprises and investment. A calculation similar to the case of
the income tax is possible, but a look at the original unaggregated IMPLAN data reveals that the
enterprise portion of the CIT captures 0.6% of the total CIT paid to the state and represents
0.0026% of the aggregated column total. Furthermore, all portions of this aggregation are
estimated by IMPLAN rather than retrieved from primary data sources. Finally, this portion of
33
the SAM plays a small role in the variables we are using to evaluate the policy simulations and
serves primarily to facilitate the investment and savings closure conditions. The enterprise
portion of the CIT is being ignored because: it is effectively inconsequential to the CIT; it is
almost lost in the column aggregation; it is based on an estimated IMPLAN SAM; and it would
require modification of numerous closure rules. Thus, we may declare a variable:
( ))
( ) (9)
where,
tyk = tax rate on capital factor k,
SAM (SG, K) = transfer from capital K to the state government SG,
SAM (Total, K) = total expenditure of capital K.
The value of this parameter is controlled in the counterfactual simulation.
The sales tax is found in two places: as a portion of the transfers from activities to
indirect taxes and in transfers from commodities to the state. Transfers to the state from
commodities are pure sales taxes. Transfers from activities to indirect taxes include state and
local sales taxes, property taxes, motor vehicle taxes, payments for licenses and fees, and
others. In this case we are not concerned primarily with the exact amount of revenue raised by
the sales taxes separate from the other indirect taxes. Rather, we are interested in the amount
of additional tax revenue raised by an increase in the sales tax rates in certain industries, as well
as the effects of these tax increases along with other tax changes on the general equilibrium
itself. Because our focus is broad enough, we can use the variable tq found in the equation for
34
household consumption of market commodities as seen in Equation 1:
( ∑ ( )
) ( ( ))
(10)
where,
QHc,h = the quantity consumed of commodity c by household h,
ϒc,h = the subsistence level of consumption of market commodity c for household h,
β c,h = the marginal share of consumption spent on market commodity c for household h,
EHh = net income of taxes, savings, and transfers,
PC = the composite price of commodity c,
tqc = the sales tax on commodity c.
This approach will allow us to determine in the model the change in revenue from the
imposition of an additional tax on commodities by industry. 6 The use of this variable is simple,
but the calculation of this variable is not. Because we wish to expand the tax to previously
untaxed sectors, while in the aggregation we have some untaxed industries combined with
taxed industries, the estimation of this tax change must include sales taxes already paid. This
can only be roughly approximated to give us an average sales tax rate on final goods. The
calculation is detailed in the simulations section.
6 Georgia also has local sales taxes that often equal the state’s rate. In the model, these would be included in the
transfers from activities to indirect taxes, allocated to the state, and then distributed to the local governments for education and investment. The amount transferred to the local governments, those that don’t collect direct taxes, for education and investment is approximately equal to the total transfers from activities to indirect taxes.
35
The final tax is the tax on intermediate inputs. There is not enough information in the
SAM to isolate this tax, declare a parameter, and control it in the counterfactual. However,
again we can ignore our inability to isolate the tax, its revenue, and its rate because we are
primarily concerned with the effects of its change. In this instance, we allow the price of
intermediate goods to be tax inclusive in the baseline. We assume the tax on inputs is zero in
the baseline and alter it in the counterfactual to simulate the reforms. Here, we include a tax on
intermediate inputs, tqinp, when calculating the quantity of intermediate use of commodity c
by activity a:
( )
( )
(11)
where,
QINTc,a = intermediate use of commodity c by activity a,
SAM (C, A) = marketed outputs of commodity c produced by activity a,
PQOc = composite commodity price of commodity c,
tqinpa,c = tax on activity a used to produce commodity c.
This variable, tqinp, also appears in the calculation of government revenue and value added
price. The Leontief coefficients in the production function are unaffected by the tax on inputs.
36
D Trade
Trade is modeled between Georgia, the RUS, and the ROW. Each region is a monopoly
provider of their differentiated exports. Trade plays a minor role in this model as the aggregate
effects of Georgia’s policy change has little effect on the value of the dollar and Georgia uses
the same currency as its domestic trading partners. For this reason, the exchange rate is held
constant while the ROW and the RUS savings are allowed to adjust. Equation 12 gives the trade
calculation for the current account balance:
∑
∑
∑
∑
(12)
where,
PWMc = import price of commodity c,
QMc = quantity of imported commodity c,
TFAf = total factor transfers to ROW and RUS,
PWEc = export price of commodity c,
QEc = quantity of exported commodity c,
TIAi = transfers from ROW and RUS to institution i,
FSAV = foreign savings.
In our closure rules, with foreign savings endogenous, the exchange rate is normalized and held
constant while foreign savings adjusts. The result is that the sum of non-government savings,
37
government savings, and foreign savings is equal to the sum of fixed investment and the stock
change.
V. CALIBRATION
A. Tax Revenue
From the SAM we derive the total amount of revenue collected by the state through
taxes on commodities, activities, households, and capital. In the table below, we show the
allocation we take from IMPLAN compared to the FY 2010 Governor’s Budget Report (in
millions of dollars). Note that we divide taxes within our SAM into the personal income tax
(PIT), the corporate income tax (CIT), and the sales tax. The category ‘other’ refers to the taxes
on motor fuels, insurance premiums, vehicle licenses, alcohol, tobacco, and property. Within
IMPLAN, we cannot identify exactly where each of these is allocated and in what proportion.
Furthermore, we are only concerned with revenue that corresponds to the state’s taxes rather
than federal and local taxes, so those tax revenues have been excluded from Table 2.
TABLE 2 – Data extract and budget comparison (in millions of $)
IMPLAN FY 2010
PIT $9,886 $7,016
Sales and Use $4,083 $4,864
CIT $ 740 $ 684
Other na $1,894
Total $14,709 $14,460
38
Because IMPLAN aggregates taxes together while our SAM contains divisions of taxes
collected by the state for both the state and localities and from numerous taxes, our numbers
do not exactly match the FY 2010 numbers from the Georgia budget. However, the totals are
very close, with our IMPLAN data giving us a total revenue number 1.7% higher than the FY2010
budget numbers.
B. Parameters
TABLE 3 – Parameter values
Parameter Value
Elasticity of substitution for production between labor and capital .9
Elasticity of substitution (Armington) between Georgia output and imports 2
Elasticity of substitution (transformation) between Georgia and foreign demand 2
Elasticity of substitution (transformation) between ROW and RUS for exports 1.35
Elasticity of substitution (Armington) between ROW imports and RUS imports 1.35
Income elasticity 1
Investment on commodities elasticity 1
Consumption flexibility -1
Investment demand flexibility -1
Supply elasticity for labor 2
Supply elasticity for capital .5
Note that the elasticity of substitution between labor and capital in the production
function is set to near unity so the CES functional form closely resembles Cobb-Douglas. The
elasticities of substitution between Georgia and its trading partners—the ROW and RUS—are
set higher than those between the ROW and RUS. This decision was made to simulate the
greater degree of openness found in regional economies than found in national or international
39
economies.7 The income elasticity and investment on commodities elasticity parameters were
taken from the recommended values of Holland, Stodick, and Davadoss (2004). The flexibility
parameters are used to set the subsistence level of consumption of investment. Having these
set equal to -1 allows for no minimum level. The supply elasticities of labor and capital reflect
that labor is more easily substituted than capital. Sensitivity analysis was conducted by altering
functional forms and parameter values with no substantive change in results.
VI. SPECIAL COUNCIL SIMULATIONS
We simulate the general tax reforms recommended by the Special Council, which
include a reduced and flat PIT of 4%, a reduced CIT of 4%, and an expansion of the 4% state
sales tax to previously untaxed food and services. In addition to these three changes, the
Special Council called for an expansion of the sales tax on inputs for the movie industry, health
services, and education services. In the case of the movie industry, this was originally done to
attract the industry to Georgia. For health and education, these were not taxed for political
reasons: when consumed as final goods, they enjoy exemptions from sales taxes on inputs.
Because the sales tax on these as final goods does not apply, tax cascading cannot occur, so it
was recommended that the exemption be eliminated. Finally, the Special Council
recommended new tax exemptions on energy inputs for agriculture, mining, and manufacturing.
7 See the average US Armington elasticities in Galloway, McDaniel, and Rivera (2003) as well as Thaiprasert, Faulk,
and Hicks (2011) for these figures use in an IMPLAN data based CGE model.
40
The PIT in the model is given by the ratio of direct taxes to total expenditures:
( ))
( ) (13)
We are interested in the ratio of direct taxes to labor income, which cannot be directly
controlled in the model. To determine the change here, we first look at the original statutory
tax code and the proposed tax code, and calculate the average tax rate for each household
under the original tax code and the proposed new code. This calculation was done for both the
original tax code and the proposed reform for single filers and filers who are married with two
children (M2K). These were then averaged to determine the ratio of the average tax rate under
the reform to the original tax code. This gives us the ratio that we can apply to ratio of the
direct tax as a proportion of labor income. We applied this ratio to our variable ty and checked
that the desired effect was achieved. The difference between the ratio of income tax rate in
the simulation to the income tax rate in the baseline, and the ratio of the income tax rate in the
reform to the income tax rate under the previous statutes is nearly identical8. This reveals that
the change in income tax we are simulating is very similar in size and scope to the proposed
reform.
Also of note regarding the direct tax rates by income group are some discrepancies in
the SAM regarding the average tax rates and the income groups. Note in Table 4 that the
calculated income per capita does not fall in the income range as aggregated in the extraction
for household groups 2, 6, 7 and 8. The disparity is largest for household 2.
8 Refer to Table A1 in the appendix for calculations of the differences between the alternative tax treatments.
41
Regarding the tax rates, they are unusually high for household group 1 and unusually
low for household group 2. This could be because the poorest group is primarily single filers
with few deductions, while household group 2 has a higher proportion of married households
with children and thus more deductions. Regardless, we apply the tax reform as calculated to
the tax rates as extracted. These unusual IMPLAN tax rates for low income groups could cause
problems if adjustments are not made relative to the baseline.
TABLE 4 – Household tax rates and income per capita (in dollars)
HHD1 HHD2 HHD3 HHD4 HHD5 HHD6 HHD7 HHD8 HHD9
Income range 0-10k 10-15k 15-25k 25-35k 35-50k 50-75k 75-100k 100-150k 150k+
Baseline tax rate 5.20% 1.35% 3.40% 4.03% 3.96% 4.26% 4.78% 4.96% 7.24%
Simulated tax rate 3.27% 1.00% 3.29% 3.96% 3.32% 3.28% 3.54% 3.54% 4.95%
Income per capita $3,458 $6,335 $16,010 $28,547 $44,748 $76,525 $103,929 $156,262 $376,941
For the expansion of the sales tax to food and services, we look at three industries: food,
other services, and information services. To simulate this tax reform we must increase the sales
tax rate from its current rate to 4%. In each of these industries, some amount of indirect taxes
is paid, so we must account for this as we calculate the additional tax. To do this, we first
calculate the amount of revenue to be raised by a 4% tax on final consumption. We then
subtract out the indirect taxes already paid. This is an approximation because we only have the
breakdown of indirect taxes as an aggregation of all industries. However, from the IMPLAN
breakdown we know that the proportion of sales taxes paid to the state as a proportion of total
indirect taxes is 21%. Using this percentage, we can estimate the amount of sales taxes already
paid as a proportion of sales taxes that would be raised by a 4% sales tax, and then calculate
42
the additional tax needed to raise that revenue. This simulates an expansion of the sales tax so
that the average sales tax rate is 4%.
We alter the tax rate on intermediate inputs to simulate the elimination or expansion of
taxes on inputs used by specific industries. For the elimination of the tax exemption for movies,
health services, and education services, we increase the taxes paid on activities used to produce
the industry’s commodities. For movies, the Special Council recommends the sales tax be
applied to purchases of “production equipment by film producers and film production
companies used to produce films for nationwide distribution.” To simulate this, we introduce a
4% intermediate input tax to movie industry activities used in the production of movie
commodities. For health and education services, this tax applies to all activities that produce
health and education services. To introduce new sales tax exemptions for energy used by the
agriculture, mining, and manufacturing industries, we introduce a negative intermediate input
tax on the use of utilities by these industries. This last change to the intermediate input price of
utilities for agriculture, mining, and manufacturing is an approximation of the actual change, as
the utilities industry is not exclusively energy.
Three simulations are done relative to the calibrated baseline. The three simulations
differ according to their treatment of labor and capital mobility and the endowments. In the
long run, both capital and labor are mobile and the endowments are variable. There are two
short run simulations. In the shortest run, labor is mobile between industries, but the total
endowment is fixed. Capital, in the shortest run, is industry specific and fixed by activity.
Relaxing the capital restrictions from the shortest run, we get a short run simulation, with labor
43
again mobile between industries with a fixed total endowment, but capital is mobile with a
fixed total state endowment.
VII. SPECIAL COUNCIL SIMULATION RESULTS
We compare the simulations to the baseline scenario and one another by looking at the
effects on the state’s economy and budget, households, and industries.
A. State Results
There are four ways we can look at the effects of tax reform on the state. These are
labeled in Table 4 as GDP-1, GDP-2, GDP-3, and GDP-4. GDP-1 is the wage and capital bill at
factor cost. GDP-2 is the traditional consumption, investment, government, and net exports
calculated at market prices. GDP-3 is the value added given by the wage and capital bill plus
indirect taxes. GDP-4 is given by the total activity minus intermediate costs. Note that in the
baseline, GDP-1 and GDP-4 are equal, and GDP-2 and GDP-3 are equal. In the counterfactual
simulations, we see that they differ as the numeraire prices set to unity in the baseline change
in the simulations.
In Table 5, we see the results given in millions of dollars. The GDP changes are all
positive for short run and long run simulations. The increase is more than twice as large in the
long run as in either of the short run simulations. We see the GDP increases slightly in the
shortest run than in the short run, likely due to the slightly higher prices that result in this
44
simulation. We can interpret this result by saying that, within the model, GDP for the state will
increase by a bit less than one tenth of one percent in the long run, given this tax reform.
However, results of a magnitude this small should be interpreted with restraint. Because the
changes are so small, a more precise interpretation may be that this revenue neutral tax reform
will have a negligible impact on the state’s GDP.
TABLE 5 – Baseline and simulated state GDP (in millions of dollars)
Base Calculated Difference Percent Change
Long Run
GDP-1 393,829 394,117 287 0.07%
GDP-2 423,435 423,775 339 0.08%
GDP-3 423,435 423,785 350 0.08%
GDP-4 393,829 394,107 277 0.07%
Short Run
GDP-1 393,829 393,941 112 0.03%
GDP-2 423,435 423,583 147 0.04%
GDP-3 423,435 423,593 157 0.04%
GDP-4 393,829 393,931 102 0.03%
Shortest Run
GDP-1 393,829 393,954 125 0.03%
GDP-2 423,435 423,589 154 0.04%
GDP-3 423,435 423,598 163 0.04%
GDP-4 393,829 393,944 115 0.03%
Regarding the state government’s budget, Table 6 breaks out the changes to revenue
from various sources of tax revenue. Note that the revenue from intermediate taxes is set to
zero in the baseline, although the state does collect this tax. The currently collected
intermediate tax is found within indirect business taxes, which are not included here as they are
transferred to the localities. Also not included here are transfers to the state from the federal
45
government which remain constant. The intermediate row and the sales tax row are separated
in the simulation SAM, so that their sums give us the changes caused by our tax adjustments.
TABLE 6 – Baseline and simulation tax revenue by source (in millions of dollars)
Baseline Long Run Short Run Shortest Run
Calculated Difference Calculated Difference Calculated Difference
PIT 9,886 7,435 -2,451 7,429 -2,457 7,430 -2,457
CIT 816 564 -252 564 -252 564 -252
Sales 4,083 6,727 2,644 6,724 2,641 6,727 2,644
Intermediate 0 -10 -10 -10 -10 -9 -9
Total 14,785 14,716 -70 14,707 -78 14,712 -74
Here we see that the state collects less total taxes, but the reform is almost budget
neutral, as the largest change in any of the simulations is $78 million dollars less than the
original tax revenue of $14.79 billion. We also see that the change adds slightly more revenue
in the long run model than in either of the short run models.
B. Household Results
We calculate utilities and equivalent variations for each household group under each of
the three simulations. We see in Tables 7 and 8 that the effect of the tax reform is minimal. This
is expected, as the state tax burden represents a fraction of the total tax burden. With the
exception of household 1 in the long run, we see that the effects on utilities are negative for
households 1 through 5, those making less than $50,000, and positive for households 6 through
9, those making more than $50,000.
46
TABLE 7 – Baseline and simulation utility by household
Baseline Long Run Short Run Shortest Run
Calculated % Difference Calculated % Difference Calculated % Difference
HHD1 7.359 7.361 0.021 7.358 (0.012) 7.359 (0.010)
HHD2 6.912 6.912 (0.006) 6.910 (0.038) 6.910 (0.037)
HHD3 7.798 7.796 (0.025) 7.794 (0.047) 7.794 (0.046)
HHD4 8.032 8.028 (0.048) 8.027 (0.064) 8.027 (0.063)
HHD5 8.523 8.522 (0.012) 8.521 (0.025) 8.521 (0.024)
HHD6 9.043 9.045 0.016 9.044 0.008 9.044 0.009
HHD7 8.653 8.656 0.038 8.656 0.030 8.656 0.032
HHD8 8.657 8.662 0.056 8.662 0.050 8.662 0.051
HHD9 9.034 9.040 0.070 9.039 0.063 9.039 0.064
We can look at Table 8 for the equivalent variations, which show us the amount of
money represented by the change in utility. Again, we see the total change for households is
small and the net effect of the tax change is regressive in nature. This is a result of the relatively
smaller decrease in PIT for middle-income people and the relatively larger proportion of their
expenditure that is used to consume goods, like food, which are now taxed.
TABLE 8 – Equivalent variation by household source (in dollars)
Long Run Short Run Shortest Run
HHD1 16.55 (9.31) (7.91)
HHD2 (2.87) (18.48) (17.94)
HHD3 (33.50) (63.33) (62.70)
HHD4 (83.38) (112.56) (110.99)
HHD5 (38.62) (78.75) (76.56)
HHD6 85.04 40.47 45.90
HHD7 134.06 105.20 109.48
HHD8 193.05 171.17 174.62
HHD9 341.28 304.30 310.98
47
C. Industry Results
Industries are aggregated into 15 categories, as labeled in Table 89. The industries give
us labels for both activities and commodities. Activities and commodities are differentiated
using by an ‘A’ or ‘C’, respectively, as seen in Tables 10 and 11.
TABLE 9 – Aggregated industry labeling key
Aggregation Label Aggregation Description
AFFH Agriculture, forestry, fishing, and hunting
MIN Mining
MANUF Manufacturing
MEDIMAN Medical equipment manufacturing
UTIL Utilities
FOOD Groceries for home consumption
INFOSER Information services
HEEDSER Health and education services
OTHSER All other services
CONST Construction
WHOLTRAD Wholesale trade
RETTRAD Retail trade
TRAN Transportation
MOVIE Motion picture and music video production
GOVSPC Government services
In Tables 10 and 11 we have industrial output by commodity and by activity. In analyzing
these results, we see the importance of the commodity and activity structure where a
commodity can be produced by any activity and any activity can produce any commodity. For
9 Table 8, identical to Table 1, is placed here for the reader’s convenience
48
example, a farm can produce both agriculture and wholesale trade commodities if that firm is
vertically integrated.
Table 10 shows the value of the total amount of a commodity that is produced
regardless of which activity produced it. Table 11 shows the value of all of an activity’s output
regardless of which commodities it is producing. Here we can see that total industrial activity
and commodity production are increased by between a fifth and a quarter of one percent in the
long run. The biggest growth sectors are utilities, retail trade, and health and education
services. Note that health and education services had increased taxes on inputs; taxes on
utilities decreased when used as an input across agriculture, mining, and manufacturing; and
retail trade was unaffected. The hardest hit industries by the reform are food, other services,
and movies, which all had tax increases imposed upon them. Total output increases as we
progress from the shortest run to the long run, and as capital and labor adjust.
49
TABLE 10 – Output by commodity (in millions of dollars)
Baseline Long Run Short Run Shortest Run
Calculated % Difference Calculated % Difference Calculated % Difference
AFFH-C 10,038 10,033 (0.05) 10,036 (0.02) 10,037 (0.01)
MIN-C 1,781 1,786 0.29 1,787 0.32 1,781 (0.01)
MANUF-C 103,260 103,371 0.11 103,361 0.10 103,233 (0.03)
MEDIMAN-C 5,583 5,594 0.20 5,595 0.22 5,584 0.03
UTIL-C 14,450 14,514 0.45 14,505 0.38 14,516 0.46
FOOD-C 45,166 44,948 (0.48) 44,959 (0.46) 45,049 (0.26)
INFOSER-C 38,675 38,776 0.26 38,765 0.23 38,730 0.14
HEEDSER-C 53,522 54,095 1.07 54,040 0.97 54,035 0.96
OTHSER-C 245,030 244,113 (0.37) 243,963 (0.44) 243,969 (0.43)
CONST-C 33,797 33,795 (0.01) 33,790 (0.02) 33,795 (0.01)
WHOLTRAD-C 38,985 39,152 0.43 39,136 0.39 39,121 0.35
RETTRAD-C 32,450 32,790 1.05 32,763 0.96 32,762 0.96
TRAN-C 26,788 26,867 0.29 26,855 0.25 26,834 0.17
MOVIE-C 1,227 1,225 (0.16) 1,225 (0.19) 1,226 (0.05)
GOVSPC-C 79,687 79,820 0.17 79,800 0.14 79,803 0.15
Average
0.22
0.19
0.16
TABLE 11 – Output by activity (in millions of dollars)
Baseline Long Run Short Run Shortest Run
Calculated % Difference Calculated % Difference Calculated % Difference
AFFH-A 9,622 9,618 (0.04) 9,622 0.01 9,619 (0.04)
MIN-A 1,734 1,739 0.33 1,740 0.38 1,733 (0.03)
MANUF-A 102,187 102,304 0.11 102,294 0.10 102,127 (0.06)
MEDIMAN-A 5,475 5,486 0.21 5,488 0.24 5,474 (0.01)
UTIL-A 11,745 11,900 1.32 11,896 1.29 11,785 0.34
FOOD-A 45,098 44,900 (0.44) 44,919 (0.40) 45,039 (0.13)
INFOSER-A 44,652 44,764 0.25 44,760 0.24 44,659 0.02
HEEDSER-A 49,882 50,415 1.07 50,349 0.94 50,297 0.83
OTHSER-A 236,884 235,975 (0.38) 235,859 (0.43) 236,267 (0.26)
CONST-A 33,797 33,794 (0.01) 33,788 (0.03) 33,784 (0.04)
WHOLTRAD-A 38,985 39,147 0.42 39,128 0.37 39,061 0.19
RETTRAD-A 32,246 32,577 1.03 32,544 0.92 32,509 0.82
TRAN-A 26,197 26,270 0.28 26,257 0.23 26,219 0.08
MOVIE-A 1,214 1,203 (0.93) 1,202 (0.95) 1,208 (0.51)
GOVSPC-A 77,069 77,162 0.12 77,112 0.06 77,097 0.04
Average 0.22 0.20 0.08
50
VIII. INCOME TAX ELIMINATION SIMULATIONS
An interesting extension is available with our current model. One feasible alternative to
the Special Council’s proposed tax reform is a complete elimination of personal and corporate
income taxes, coupled with the replacement of this lost revenue with an expanded uniform
sales tax. With the previous simulation, we instituted the recommendations as closely as
possible to the Special Council’s proposal. In this simulation, we eliminate the state income tax
on individuals and corporations, and then increase a uniform sales tax rate on all final goods
until we achieve approximate revenue neutrality. This simulation is hypothetical and done
purely for comparison purposes. However, Georgia’s two neighboring states, Florida and
Tennessee, have tax codes with no income taxes, and as that is the general direction of the
Special Council’s reform proposal, it can provide guidance as we think about tax changes.
The specifications of this tax change include a state tax rate of zero on personal and
corporate income. We also expand the sales tax to groceries for home consumption and
previously untaxed services, so that the state sales tax rate is uniform across goods. We also,
rather arbitrarily, included the Special Council’s recommendations regarding tax deductions.
Thus, we remove the favored tax treatment of the movie industry, and include deductions on
intermediate utilities used in agriculture, mining, and manufacturing. While the inclusion of
these structural changes could have been left out of this application, they were justified to
maintain fairness, and for tax cascading reasons in the previous simulation. We also wanted to
focus this comparison on the drastic changes in sales and income taxes, without clouding the
analysis by their removal in this iteration. With this new simulation of a total replacement of
51
the personal and corporate income taxes with an increased and expanded sales tax, we again
look at the effects on the state, households, and industries.
In these simulations, the baseline of the state remains unchanged from what it was in
the Special Council simulations. The methodology for imposing constraints on the model in this
application has not changed from what was described above. The change to the treatment of
intermediate goods in the movie, agriculture, mining, and manufacturing industries is
unchanged. For both the corporate and personal income tax changes, we imposed a rate of
zero.
For the sales tax rate, we used our previous work in the expansion of the sales tax to
groceries and services. We then incrementally increased the now uniform tax rate until we
achieved approximate revenue neutrality. This process revealed that revenue neutrality was
reached with a sales tax rate of 9.2%. As this would be an unlikely tax rate for a state, we set
this at 9% and allowed for an approximately revenue neutral tax change with a net decrease in
revenue of 4%.
TABLE 12 – Sales tax revenue by treatment (in millions of dollars)
Treatment Sales Tax Rate Exclusions Revenue Revenue ÷ sales tax rate
Baseline 4% Food and services $4,083 $1,021
Special Council 4% None $6,726 $1,682
Income Tax Elimination 9% None $14,300 $1,589
Table 12 shows the average revenue collected by the sales tax across labor and capital
closure rules by treatment. Here we see the magnitude of the tax expansion to food and
services relative to the baseline, and to the expansion coupled with an increase in tax rate from
52
4% to 9%. Also note that the amount of revenue per percentage point of sales tax decreases
from $1,682 to $1,589 as we increase the rate from 4% to 9%. This result is what we would
expect, given consumers’ ability to substitute for imports as a result of our Armington
elasticities. The increase from the baseline to the treatments is a result of the expanded base.
53
IX. INCOME TAX ELIMINATION RESULTS
A. State Results
In Table 13 we have the same four GDP measures introduced in the previous simulation.
TABLE 13 – Baseline and simulated state GDP (in millions of dollars)
Base Calculated Difference Percent Change
Long Run
GDP-1 393,829 393,414 -416 -0.11
GDP-2 423,435 423,288 -147 -0.03
GDP-3 423,435 423,233 -202 -0.05
GDP-4 393,829 393,468 -361 -0.09
Short Run
GDP-1 393,829 393,446 -383 -0.10
GDP-2 423,435 423,323 -113 -0.03
GDP-3 423,435 423,268 -168 -0.04
GDP-4 393,829 393,501 -328 -0.08
Shortest Run
GDP-1 393,829 393,075 -754 -0.19
GDP-2 423,435 422,919 -517 -0.12
GDP-3 423,435 422,862 -573 -0.14
GDP-4 393,829 393,132 -698 -0.18
In this case, we get results of similar magnitude, but of an opposite sign. Note that in all
three closure options simulating long run, short run, and shortest run, the impact on the GDP is
negative and ranges from -0.19 in the shortest run, to -0.03 in the short and long run
54
treatments. The average impact in the shortest run is -0.1575. The average impacts in the two
treatments where labor and capital have greater mobility are -0.0625 in the short run and -0.07
in the long run. As with the interpretation of the changes that took place in the previous
simulation, the most important finding here is that the magnitude is in a range between -15 and
-7 basis points. The proper interpretation may not be one that emphasizes the negative values,
but rather the negligible magnitude.
Table 14 shows revenue collected, by source, by the state government. Here we see the
absence of personal and corporate income taxes, and the considerably larger amount of
revenue generated by the sales tax to compensate. The intermediate tax revenue is essentially
unchanged from the previous treatment, as those policies are again imposed here. Also, the
figures below show approximately 90% of the net decrease in revenue. The remainder of the 4%
decrease in revenue is found in a lower factor income as a result of the now lower market
clearing rental rate of capital.
TABLE 14 – Baseline and simulation tax revenue by source (in millions of dollars)
Baseline Long Run Short Run Shortest Run
Calculated Difference Calculated Difference Calculated Difference
PIT 9,886 0 -9,886 0 -9,886 0 -9,886
CIT 740 0 -816 0 -816 0 -816
Sales 4,083 14,308 10,225 14,310 10,227 14,283 10,200
Intermediate 0 -11 -11 -11 -11 -9 -9
Total 14,785 14,297 -488 14,299 -486 14,274 -511
55
B. Household Results
In our household results, we see the model confirm economic intuition. We expect an
elimination of the income tax to mostly benefit those with the highest prior tax rate. We would
further expect the expansion and increase of the sales tax to harm those for whom taxable
consumption is the largest proportion of their income. These results are found in results for
both utilities and equivalent variation.
In our utilities measures, found in Table 15, we see that the poorest households are hit
hardest by the reform, with a decreased utility between -1.6% in the short run, and -1.5% in the
long run. We see that utility changes are negative for all groups with annual incomes below the
$50,000 cutoff. For groups of annual incomes of $50,000 or more, we see utility changes near
zero or positive reaching a high for the richest household group with an increase in utility of
approximately .3% in the long run.10
10 Households 1-5 have incomes less than $50,000. For more details, refer to Table 4,
56
TABLE 15 – Baseline and simulation utility by household
Baseline Long Run Short Run Shortest Run
Calculated % Difference Calculated % Difference Calculated % Difference
HHD1 7.359 7.246 (1.531) 7.246 (1.532) 7.240 (1.617)
HHD2 6.912 6.816 (1.384) 6.816 (1.385) 6.809 (1.490)
HHD3 7.798 7.734 (0.825) 7.734 (0.826) 7.727 (0.910)
HHD4 8.032 7.993 (0.487) 7.993 (0.488) 7.990 (0.523)
HHD5 8.523 8.503 (0.240) 8.502 (0.241) 8.490 (0.385)
HHD6 9.043 9.044 0.015 9.044 0.014 9.043 0.002
HHD7 8.653 8.654 0.008 8.654 0.007 8.653 (0.002)
HHD8 8.657 8.670 0.147 8.670 0.146 8.668 0.130
HHD9 9.034 9.061 0.298 9.061 0.297 9.060 0.288
In Table 16, we see that the equivalent variation for the average household in each
income group agrees with our utility findings. The equivalent variations for the poorest five
income groups with incomes below $50,000 per year all show negative equivalent variations,
with the lowest being $543 for the poorest household group. Equivalent variations for the
richest four households are noticeably higher, with a low of -$7 and a high of $1,377. As with
utilities and in our other findings, we see very little difference between our short run and long
run closures, but both are uniformly higher than the shortest run treatment.
The magnitude of the equivalent variation gives an approximate scale to the utility
findings. It is true that, in the simulation, household income net of taxes increases for each
household, but in the after tax price of final goods, the increase ranges between 5% for
previously taxed goods, and approximately 9% for previously untaxed goods. These increases
hit the poorest households hardest, as they consume a much higher proportion of income. The
-$514 value for the poorest group in the long run represents approximately 15% of their taxable
income. This is even higher than the new increased tax rate. Obviously, if 100% of their income
were now spent on the goods with the highest tax increase, we should not expect an equivalent
57
variation proportionally above the tax rate. This proportion is high because transfers from
government, other households, and a small amount of capital income increases the total
resources available for expenditure above the calculations used for income, subject to the
personal income tax. It is also notable that the magnitude for the richest household group is
less than their savings from the income tax elimination. Their proportion is less than a single
percent of their total taxable income. This may be a result of the relatively lower return to
capital, which represents approximately 40% of their income.
TABLE 16 – Equivalent variation by household (in dollars)
Long Run Short Run Shortest Run
HHD1 (514) (515) (543)
HHD2 (283) (283) (304)
HHD3 (460) (461) (507)
HHD4 (348) (348) (373)
HHD5 (304) (306) (488)
HHD6 82 77 11
HHD7 27 23 (7)
HHD8 499 496 442
HHD9 1,377 1,366 1,339
These household findings could be an understatement of the regressivity in other ways
as well. We might expect richer households to have a greater ability to avoid taxes by importing
goods from other regions. This would require justification, but variation in the Armington
elasticities between income groups could exacerbate the regressivity. Higher sales tax might
harder hit industries such as tourism, which hire many lower-wage workers. Also, the
equivalent variation for household 1 (HHD1) may be understated because that household had a
high initial income tax rate (shown in Table 4) and thus received unusually high income tax
58
relief with its elimination. Also not included in this model are any feedback effects on federal
income taxes or transfers. For example, high earners would now be transferring more to the
federal government because they are deducting less for state income taxes paid; for the same
reason, low earners might receive less in transfers as a result of higher federal income tax
liability.
We are reluctant to draw more refined conclusions from the magnitude of the
household effects for low income groups due to peculiarities in the IMPLAN extract for these
households, especially groups 1 and 2. Also, the magnitudes for low earners are particularly
acute, while those for the high income earners are particularly weak. Finally, corner solution
models with tax rates of zero can present problems that are unresolvable with a model
designed for general rather than specific purposes. Regardless of our trepidation regarding
magnitudes, the regressivity of the reform is clear.
59
C. Industry Results
In our industry results for this treatment we use the labels found in Table 17, which are
the same as those listed in Tables 1 and 9.
TABLE 17 – Aggregated industry labeling key
Aggregation Label Aggregation Description
AFFH Agriculture, forestry, fishing, and hunting
MIN Mining
MANUF Manufacturing
MEDIMAN Medical equipment manufacturing
UTIL Utilities
FOOD Groceries for home consumption
INFOSER Information services
HEEDSER Health and education services
OTHSER All other services
CONST Construction
WHOLTRAD Wholesale trade
RETTRAD Retail trade
TRAN Transportation
MOVIE Motion picture and music video production
GOVSPC Government services
There is an important caveat to address in our industry results: our decision to impose a
9% rather than a 9.2% sales tax reveals itself here in our findings for government services. The
entire decrease of 4% of the budget is not located here, as the remainder is spread out over
transfers to households and other levels of government, but that negative value is a result of
our balanced budget closure rule and our decision to impose a sales tax without decimals. The
remaining results are endogenous and reflect the new equilibrium resulting from the policy
changes.
60
TABLE 18 – Output by commodity (in millions of dollars)
Baseline Long Run Short Run Shortest Run
Calculated % Difference Calculated % Difference Calculated % Difference
AFFH-C 9,622 10,074 0.36 10,078 0.40 10,046 0.08
MIN-C 1,734 1,791 0.54 1,791 0.57 1,782 0.04
MANUF-C 102,187 103,513 0.24 103,515 0.25 103,367 0.10
MEDIMAN-C 5,475 5,624 0.74 5,626 0.77 5,594 0.20
UTIL-C 11,745 14,750 2.08 14,753 2.10 14,500 0.35
FOOD-C 45,098 45,186 0.04 45,202 0.08 45,163 (0.01)
INFOSER-C 44,652 38,547 (0.33) 38,555 (0.31) 38,644 (0.08)
HEEDSER-C 49,882 55,160 3.06 55,158 3.06 54,956 2.68
OTHSER-C 236,884 244,241 (0.32) 244,277 (0.31) 244,613 (0.17)
CONST-C 33,797 33,308 (1.45) 33,308 (1.45) 33,347 (1.33)
WHOLTRAD-C 38,985 39,307 0.83 39,308 0.83 39,206 0.57
RETTRAD-C 32,246 33,215 2.36 33,215 2.36 33,119 2.06
TRAN-C 26,197 26,999 0.78 26,999 0.78 26,931 0.53
MOVIE-C 1,214 1,199 (2.27) 1,200 (2.25) 1,216 (0.94)
GOVSPC-C 77,069 78,029 (2.08) 78,024 (2.09) 78,315 (1.72)
Average
0.31
0.32
0.16
In Tables 18 and 19, we see that there is little difference between the effects by
commodity and activity, respectively, as the values are quite similar. As in the case of the
Special Council simulation, we see negative effects on food, all newly taxed services, and the
movie industry. The exception here is in health and education, which sees robust growth. We
also note that the utilities industry, which enjoys some new tax advantages, grows fastest. The
other notable success is in retail trade, which survives the higher sales tax and apparently
thrives with the now moderately lower equilibrium wage rate. The remaining industries are
unaffected or see moderate growth. The average increase is between 0.16% in the shortest run
and 0.39% in the long run.
61
TABLE 19 – Output by activity (in millions of dollars)
Baseline Long Run Short Run Shortest Run
Calculated % Difference Calculated % Difference Calculated % Difference
AFFH-A 9,622 9,659 0.38 9,662 0.42 9,630 0.08
MIN-A 1,734 1,744 0.57 1,744 0.61 1,734 0.04
MANUF-A 102,187 102,441 0.25 102,443 0.25 102,294 0.11
MEDIMAN-A 5,475 5,516 0.76 5,518 0.79 5,486 0.21
UTIL-A 11,745 12,109 3.09 12,112 3.12 11,847 0.87
FOOD-A 45,098 45,119 0.04 45,135 0.08 45,096 (0.01)
INFOSER-A 44,652 44,504 (0.33) 44,513 (0.31) 44,616 (0.08)
HEEDSER-A 49,882 51,521 3.28 51,519 3.28 51,316 2.87
OTHSER-A 236,884 236,141 (0.31) 236,176 (0.30) 236,494 (0.16)
CONST-A 33,797 33,308 (1.45) 33,308 (1.45) 33,347 (1.33)
WHOLTRAD-A 38,985 39,307 0.83 39,308 0.83 39,206 0.57
RETTRAD-A 32,246 33,015 2.39 33,015 2.39 32,919 2.09
TRAN-A 26,197 26,420 0.85 26,421 0.85 26,350 0.59
MOVIE-A 1,214 1,186 (2.27) 1,187 (2.25) 1,203 (0.93)
GOVSPC-A 77,069 75,302 (2.29) 75,297 (2.30) 75,607 (1.90)
Average
0.39
0.40
0.20
If we remove government services from the calculation above to see the growth rates of
all industries without the artificial decrease we imposed by our inexact budget neutral tax
change, we have a higher average for the remaining 14 private sector industries. Using this
measure, the average growth rates vary from 0.29% in the shortest run for commodities, and as
much as 0.59% for activities in the short run. Table 18 shows the values for both private sector
commodity and activity growth rates and their average across treatment. The average of the
growth rates of both commodities and activities is an arbitrary measure that can be interpreted
as the growth of the general sector of the economy.
62
TABLE 20 – Average private sector growth rate of commodities and activities
Long Run Short Run Shortest Run
Commodities 0.48 0.49 0.29
Activities 0.58 0.59 0.35
Average 0.53 0.54 0.32
X. CONCLUSIONS
In both of our simulations, we see what one would expect from a modest subnational
tax reform package that is intended to be revenue neutral: modest effects. In either simulation,
the effect on gross state product is measured in basis points. In the Special Council reform, the
effect on GDP is negligibly positive, while in the income tax elimination simulation the effect is
negligibly negative. In both cases, the long run, with mobile and variable capital and labor, has
a higher GDP than in the shortest run, with activity specific capital and fixed supplies of both
labor and capital. This suggests that either reform will cause some growing pains before all
adjustments are made.
In the Special Council’s simulation, Georgia’s wealthier four household groups would be
wealthier, in the long run, by at least $85, while the poorest five groups are either unaffected or
poorer by no more than $83 dollars. In the long run, those making less than $50,000 will have
their utilities decline by less than five basis points, while those making more than $50,000 will
have their utilities increase by as much as seven basis points. These dollar figures and utility
changes are so insignificant as to be disregarded in the decision whether or not to institute the
Special Council reform.
63
The more extreme hypothetical of replacing the income tax with an expanded uniform
sales tax reveals the regressivity of the general direction of the more balanced Special Council
reform. We see the poorest lose 1.5% of their utility while the richest gain 0.3%. In dollar terms,
the poorest are hit with a negative equivalent variation greater than the proportion of the new
sales tax to their taxable income, and the richest benefit by less than 1%. Again, these effects
are in the direction we would expect in theory, and even the largest swing in utility is small
enough to perhaps be compensated for by transfers. However, the total negative equivalent
variation for the 58% of the population that has a negative value would require more in terms
of transfers than is gained by the remainder.
In the Special Council simulation, industrial output is approximately 0.25% above what it
would have been, although food, other services, and movie industrial sectors suffer. The
remaining industries are unaffected or have output as much as 1.07% higher. In the income tax
elimination model, we see more robust growth from our industries than that seen in the Special
Council simulation, with a bit more than 0.33% increase in the long run. Furthermore, this
growth would be 0.5% without inclusion of the artificially low government.
As planned, the Special Council’s tax reform is essentially budget neutral, with a shortfall
of $70 million. This would be diminished by the elimination of the sales tax holiday or other
unsimulated reforms. The income tax elimination simulation makes an arbitrary decision to
allow for a 4% decrease in the size of the state’s budget to avoid an unwieldy 9.2% sales tax.
This 4% decrease is the approximate size of the simulated corporate income tax relief.
The imposition of either tax reform is not unambiguously positive for all participants in
the economy. In the Special Council simulation, we see that the GDP and output increase, so
64
we know that this reform expands economic activity. Furthermore, the welfare gains from
those with positive equivalent variations are enough to compensate for the losses. Regarding
the efficacy of the distribution of gains in industry, those industries that had previously
benefited from exclusions in the tax code suffer when those exclusions are removed and are
treated like the rest of the sectors under the uniform code.
The income tax elimination is a more extreme adjustment and may require additional
model refinement to draw precise conclusions. However, this approach is considerably more
regressive than the Special Council’s recommendation. The effect on GDP, the budget, and
industry output are mixed, with the first being nearly unaffected, the second decreases, and the
latter increases.
Further research may include dynamics to simulate the proposed implementation over
time to tease out the adjustment path. Additional attention could also be paid to household
specific elasticities to try to get more precise measures of magnitude for the household effects.
Alternative industry aggregations could be considered to observe subsectors of industries that
are more affected by the reforms. Finally, one could include additional parameters for the
feedback effects on individuals regarding transfers received from the government or taxes paid
as a result of diminished taxable income.
65
CHAPTER THREE: A SIMULATED FUTURE OF EGYPT
Thanks to coauthors: Manal Metwaly, Nour Abdul-Razzak, and Andrew Feltenstein
I. INTRODUCTION
Computable general equilibrium (CGE) models are frequently used to simulate fiscal
policy changes, trade liberalization, or monetary policy actions to estimate the resulting effects
on economic variables such as growth, income distribution, or deficits. However, the CGE
framework also allows for inclusion of other exogenous events by altering key parameters after
calibration. In most cases, this exercise is applied to simulate some minor change to the
productivity of factors, as in the effect of a drought on an economy by decreasing the
productivity of land in the agriculture sector, for example. The application of this approach to
more drastic changes has, in the past, been limited by the unavailability of important data.
The Egyptian revolution in 2011 provides an interesting opportunity to use this feature
of CGE models to study a drastic change to an economy. Egypt’s revolution had this type of
substantial macroeconomic impact by negatively impacting key industries, balance of payments,
budget deficits, foreign direct investment, and other observable and measurable variables.
Furthermore, as the revolution took place, the Egyptian bureaucracy continued to collect data
on the economy. This paper will use pre-revolution historical data for calibration and post-
revolution data to incorporate known exogenous effects of the Egyptian revolution in a
66
dynamic general equilibrium model that endogenously generates tax avoidance behavior. We
will then simulate the current policy and a fiscal reform in post-revolution Egypt.
II. EGYPT BEFORE AND AFTER THE REVOLUTION
Over the past twenty years, Egypt has transformed its economy. Beginning in 1991 with
the Economic Reform and Structural Adjustment Program, Egypt restored macroeconomic
equilibrium by increasing the role of the private sector, relaxing price controls, reducing trade
restrictions, and taming inflation and budget deficits. Egypt instituted structural reforms in the
area of monetary policy in 2000 with abolishment of the de jure exchange rate peg. It
introduced a domestic currency overnight interbank market, and a foreign exchange interbank
market, in 2001 and 2004, respectively. In 2005, in an effort to control inflation and stabilize
prices, the Central Bank of Egypt (CBE) established a ‘corridor system’ with a ceiling and a floor
for the overnight interest rates on both lending from and deposits to the CBE. Their goals were
not met. According to the International Monetary Fund (IMF), inflation at the beginning of the
decade was under 3%, but by 2008 it had reached 20%.
Modest tax reform took place in 2001, with the general sales tax extended to the
wholesale and retail sectors. Major reforms were enacted in 2005. The top personal income
tax rate was decreased from 32% to 20%. Taxes on corporate profit were cut from 40%, to 20%
on profits under $10 million EGP, and to 25% on profits above $10 million EGP. Property tax
rates were cut from 46% to 10%. Sales taxes were standardized and cut from 46% to 30% for
luxury goods; 3% to 0% for essential goods; and a uniform rate of 10% for all non-luxury,
67
nonessential goods. Further restructuring included the consolidation of numerous tax
collection agencies into a single agency and the implementation of self-assessment of tax
liability.
Following the major fiscal and monetary reforms in 2005, Egypt’s economy tracked in a
generally positive direction. Growth rates reached as high as 7.2% before the financial crisis
triggered a low of 4.7%. Unemployment fell from a rate of 11.5% in 2005 to 9% in 2010. The
debt- to-GDP ratio fell from 103% to 70% in 2008. Inflation, however, rose from 4.7% to 20.2%
in the same period. This inflation is often cited as an instigating factor of the revolution, as the
poor were hardest hit by rising food prices.
In January of 2011, the Egyptian revolution ousted Mubarak from the Presidency of
Egypt, the office he had held for nearly 30 years. According to protestors, the revolution was
largely a political one for democracy, and against police and government corruption. We
believe that it is reasonable to assume that Egypt’s new leadership will not drastically alter the
structure of the economy by nationalizing private companies or pursuing socialist or isolationist
policies. For the purpose of our model, we assume that the largely free market economic
system in Egypt, as it existed prior to the revolution, will continue to exist.
The direct effects of the revolution on the economy were significant. During the
revolution, crime rates rose causing tourism to fall dramatically. Tourism in 2010 accounted for
13% of GDP, 11% of total employment, $14 billion in export revenue, and 22% of total exports
(WTTC, 2010). According to the Central Agency for Public Mobilization and Statistics (CAPMAS),
68
tourism fell by one third in 2011. Outside sources have put the figure as high as 70%11. After
the revolution, unemployment increased from 9% to 12.1%; real GDP growth fell from 5.1% to
1.8%; and Egypt saw its bond ratings downgraded by all three major ratings agencies (EMOP).
Business starts decreased, consumption fell, and investment decreased. Quarterly foreign
direct investment fell from an inflow of nearly $2.5 billion to an outflow of $163 million (MOED).
Regarding fiscal policy, following the revolution, leaders stipulated a 15% increase in wages and
pensions for the government sector, and the top marginal income tax rate was increased to
25%.
Billions of dollars were pledged by foreign governments to the Egyptian people in the
form of foreign aid for post-revolution recovery. Initial infusions of cash and petroleum
products from Qatar and Saudi Arabia have kept the currency afloat. The interim military
government was able to raise funds by selling Egyptian bonds, but the interest rate on a one
year bond was 16%. After the June 30, 2012, election of Mohamed Morsi as the new Egyptian
President, the IMF worked unsuccessfully with the new Egyptian government to set terms of a
$3.2 billion loan. Morsi’s first official trip was to Saudi Arabia, where he received $1.75 billion
in aid. In the first year after the revolution, foreign exchange reserves fell from $36 billion to
$15.5 billion, enough to cover three months of imports.
In November 2012, President Morsi issued a declaration immunizing his decrees from
legal challenge. The president justified this as a move to protect the work of the constitutional
11 It is difficult to use CAPMAS data for tourism because Tunisian, Libyan, and other refugees cannot be separated
from tourists.
69
assembly, but instead it set the stage for a second uprising. The following month, the
constitution was approved by referendum, despite protests, with approval of 63% of voters.
The unrest continued until a large protest, scheduled in June 2013, continued until July 3, when
Egyptian army chief general Abdel Fattah el-Sisi removed Morsi from power. The next day, el-
Sisi named Adly Mansour, a former judge, as Egypt’s interim President, but observers noted
that el-Sisi was running the country. In January 2014, a new constitution was enacted with 98%
approval. Elections for the Presidency are scheduled for late May 2014, with el-Sisi polling well
after the March 26th announcement of his candidacy.
The new Egyptian President will inherit an economy with significant problems. As a
result of a lack of security, crime, curfews, and civil unrest, GDP growth fell to an annualized
rate of 1% in the first quarter of 2014. GDP forecasts for 2014 range between 2% and 2.5%.
The deficit is currently at 13% of GDP, with a debt-to-GDP ratio of nearly 90%. Aid from Saudi
Arabia and others has kept foreign reserves above the 3 month minimum, but inflation has
increased over the last few months from a 7.6% annual rate to a monthly rate equal to an
annual inflation of more than 13%. Table 1 displays key indicators of the Egyptian economy
before and after the 2005 tax reforms and the revolution. Inflation continues to rise and
impede investment and confidence in the stability of the Egyptian economy. Furthermore,
despite efforts to reduce public spending, public domestic debt has increased to 90% compared
to an average of 50% in the Middle East and North Africa (Achy, 2010).
70
TABLE 21 – Egyptian Economy Before and After 2005 Tax Reforms and 2011 Revolution
Year Nominal GDP (Billion US$)
Real GDP (Billion EGP)
Real GDP Growth Rate
Price Index
(1999 = 100)
Inflation Rate
Unemployment Rate
Gov’t Debt (% of GDP)
2000 100 355 5.4 101.1 1.1 9.0 n/a
2001 97 367 3.5 103.4 2.3 8.8 n/a
2002 86 379 3.2 106.2 2.7 10.1 90.4
2003 81 391 3.2 110.5 4.1 11.3 102.3
2004 79 407 4.1 123.4 11.7 10.5 101.5
2005 90 425 4.5 129.2 4.7 11.5 103.3
2006 107 454 6.8 138.5 7.2 10.9 90.3
2007 130 487 7.1 150.4 8.6 9.2 80.2
2008 162 521 7.2 180.7 20.2 8.8 70.2
2009 189 546 4.7 198.7 10.0 9.5 73.0
2010 219 574 5.1 220.0 10.7 9.0 73.2
2011 236 584 1.8 245.9 11.8 12.1 76.6
2012 262 597 2.2 272.5 10.8 12.3 80.6
2013 271 609 2.1 289.5 6.2 13.0 89.5
Source: International Monetary Fund, World Economic Outlook Database, April 2014 **At an exchange rate of $1=£0.155, Egyptian Nominal GDP is approximately £1,753 in 2013.
71
III. EGYPTIAN TAX EVASION AND INFORMAL SECTOR
A. Tax Evasion
High levels of tax evasion represent serious challenges for many developing nations as
they seek to grow and expand their public services. Trade liberalization is a significant part of
many reforms in developing nations. Consequently, there is a need to increase domestic
revenues to replace revenues lost to decreased tariffs (Toye, 2000). Furthermore, if taxes are
too high or too cumbersome to pay, the threat of citizens’ migration to an informal economy
becomes more likely (Turnovsky and Basher, 2009). Thus, it is in the government’s best interest
to develop policies that creatively increase tax compliance while promoting economic growth.
Egypt is an interesting case, as it has transformed itself in the past thirty years from an
economy dominated by the public sector to a private sector and market-oriented economy
(Dobronogov et al., 2005). At the same time, Egypt has implemented a series of tax reforms.
Although many other Arab countries have instituted tax reforms and structural adjustment
programs, Egypt is the largest and most diverse non-oil based Arab economy to manage such
extensive reforms. Egypt has seen moderate growth since the implementation of structural
adjustment programs. However, following the global financial crisis of 1997 and a domestic
financial scandal in 1998, Egypt’s economic growth slowed and failed to sustain a high growth
rate. Prior to the revolution, negative global shocks continued and Egypt’s fiscal situation
deteriorated with steadily increasing budget deficits. According to the IMF, fiscal vulnerabilities
remain as Egypt’s main macroeconomic risk (“Arab Republic of Egypt…” IMF, 2010).
72
Egypt’s tax revenue-to-GDP ratio in 2010 was only 14.1%12, which is relatively low
compared to other Arab economies of similar size and nature, indicating the need to explore
policies that curb tax evasion and improve growth rates. Determining the extent to which the
tax administration in Egypt is a significant constraint on private sector development is vital. A
survey of 1000 enterprises conducted by the World Bank13 in 2004 found that tax-related issues
were ranked as severe by almost 80% of participants (“Egypt National Investment…” OECD,
2006). Tax authority officials are known to exercise significant discretion in applying rules
related to enterprise regulation and public service access. To complicate matters, tax laws are
excessively complex, thereby giving more room for the tax authority to act arbitrarily (“Egypt
Country Profile…” FEMISE, 2004). These issues have led to an underreporting of income, tax
evasion, and subsequently low revenues for the government. Although there is a dearth of
literature in this area, the few articles that have explored the cost of tax evasion in Egypt
conclude that the benefits of tax evasion exceed its costs at this time (El-din et al., 2000)14. An
analysis done by the Egyptian Center for Economic Studies found that the levels of tax evasion
in Egypt and other developing nations is a serious issue (Tohamy, 1998). With tax revenue at
14.1% of GDP in 2010, Egypt’s tax effort is relatively similar to that of the Philippines at 15.7%,
Indonesia at 15.5%, and Thailand at 15.7% (Gordon and Li, 2009). Developing nations’ tax
revenues average about 19.6% of GDP compared to 28.4% in the developed world. However,
12 Some additional revenue is obtained from Suez Canal fees that is not included in the tax to GDP ratio, but the
ratio only increases to 16.8% with its inclusion.
13 World Bank Investment Climate Survey of Egypt
14 El-din, Fawza and Refaat did not determine how tax evasion affects investment quantitatively because of the
lack of information regarding transactions costs of tax compliance and the magnitude of tax evasion.
73
Egypt’s tax-to-GDP ratio falls well below that of other neighboring Arab countries with similar
economies, such as Morocco (at 24%), Tunisia (at 21%) and Jordan (at 20%). This further
emphasizes Egypt’s weak fiscal situation, possibly as a result of high rates of tax evasion. As
with many other problems in Egypt, tax evasion was exacerbated by the revolution, when tax
revenue as a percent of GDP fell to 13.2%. This put Egypt slightly below Columbia (13.3%) and
slightly above Uganda (13%).
Although Egypt was labeled the top tax policy reformer in the world by the Doing
Business Report of 2008, substantial obstacles remain, especially for small and medium
enterprises that comprise 80% of Egypt’s GDP. A World Bank Investment Climate Survey from
2004 found that access to financing and its associated costs are the main reasons for
constrained investment and growth of small and medium enterprises. Credit in the private
sector has averaged around 40% of GDP in the past 5 years. Access to bank credit is necessary
to support growth of the private sector, and it could partially replace self-financing and the now
absent foreign direct investment. A survey conducted by the Egyptian Center for Economic
Studies found a mismatch between banks’ excess lending capacity and the volume of loans
provided to the private sector (Abdel-Kader, 2006). For a majority of firms (70%), the main
source of financing was the firm’s own funds (retained earnings).
Despite a large banking system, credit in the private sector remains concentrated
among large enterprises (“International Bank for Reconstruction…” World Bank, 2010). Over
half of the credit given to the private sector goes to less than 0.2% of bank clients, according to
the Central Bank of Egypt. The 2009 Investment Climate Assessment survey found that firm size
is the most important characteristic associated with whether or not a firm had a loan or
74
overdraft capacity. Reforms have taken place to increase access to credit, including creating a
new private credit bureau: iScore. Measures were also taken to reduce the minimum capital
required to start a business, from 50,000 Egyptian pounds to 1,000 Egyptian pounds (Alissa,
2007). Egypt has strong future growth prospects; however, the low level of private credit to
businesses and households as a percentage of GDP remains problematic (IMF, 2010).
B. Egyptian Informal Sector
With an estimated 8.2 million people in the informal sector (37% of the workforce),
there is great opportunity to broaden Egypt’s tax base (World Bank 2010). A USAID report
found that between the years 1998 to 2006, the informal economy grew at an annual rate of
5.3% (Alissa, 2007). Furthermore, USAID found that about 40-60% of the cost of doing business
is result of the cumbersome regulatory framework that promotes the informal economy.
Very few studies have explored the nature of the informal economy in Egypt. Economist
Hernando de Soto of the Institute for Liberty and Democracy led years of fieldwork and analysis
for the Egyptian government to determine how much of the economy operated “extralegally”
(de Soto, 2011). This notable study concluded that the underground economy was the nation’s
largest employer in 2004. He found that 9.6 million people worked in the extralegal sector,
while 6.8 million were employed in the private sector, and 5.9 million in the public sector.
Furthermore, 92% of Egyptians hold their property without normal legal title. Without any legal
title to assets and real estate, entrepreneurs cannot leverage these assets as collateral for loans
75
or investment capital. As a result, the majority of enterprises remain small and poor. At
the time, de Soto estimated the value of these extralegal businesses and properties (in both the
rural and urban parts of Egypt) to be $248 billion, 6 times greater than total savings and
deposits in commercial banks in Egypt, and 30 times more than the market value of the
companies registered in the Cairo Stock Exchange. The survey also identified the primary causes
for extralegality in Egypt, notably the convoluted and redundant legal requirements for the
various stages of a firm’s development. The primary cause identified for the difficulty of
expansion is the lack of access to credit. de Soto’s findings have since been corroborated by
other Egyptian studies and surveys.
The Egyptian Labor Market Survey Data for 2006 found that medium and large firms hire
nearly 25% of their workers informally and thus without a secure contract or social security
(World Bank, 2010). The 2009 Investment Climate Survey by the World Bank found that
managers surveyed ranked tax rates, and uncertainty about economic and regulatory policies as
major constraints to businesses. Furthermore, El-din et al. (2000) found that taxation has a
major impact on competitiveness and net profitability through its effect on the cost of capital in
Egypt. They showed that debt financing alleviates the effective tax burden on projects and
therefore provides an additional incentive to borrow. However, this means of financing new
investments is not easily accessible to non-corporate businesses or small or medium
enterprises, as they are unable to provide necessary collateral to the banks. As a result, small
firms are left with a higher burden of taxation than are large corporations.
Given these findings, we see that even after the tax reforms of 2005, evasion and
informality remain key impediments for growth. Many Arab countries have sought taxation
76
reform through improvements in tax administration and institutional reform. Furthermore,
many countries have instituted full-fledged value added taxes (VAT), with Morocco having one
of the highest rates in the region. Other Middle East and North African (MENA) tax reforms
include lowering corporate income tax rates and implementing online systems (Paying Taxes,
2010). Jordan recently simplified tax forms and introduced an electronic payment system. In
2009, Tunisia mandated that all companies with a high turnover use the online tax system. Both
Djibouti and Iran have recently replaced their sales tax with a new value added tax.
Comparisons within the MENA region have shown the detrimental effects of widespread
tax evasion and informal economies similar to those found in Egypt. Given the difficult
economic times, the revolution, and the inefficient tax system, new policies are needed to
stimulate growth, increase revenues, and increase private sector access to credit. As explained
below, this model accommodates the large amount of tax evasion and informality as it
correlates the underground economy to tax rates and the need to access the banking system.
IV. GENERAL EQUILIBRIUM SPECIFICATION
In this section we develop the formal structure of the dynamic general equilibrium
model. This is a dynamic model that endogenously generates tax avoidance behavior by firms.
We will detail the model specifications regarding dynamics, production, banking, consumption,
government, and the foreign sector.
77
A. Dynamics
Our model is a sequentially dynamic model15 and thus has n discrete time periods. All
agents optimize in each period over a 2 period time horizon. That is, in period t they optimize
given prices for periods t and 1t with expectations for prices for the future after period 1t .
When period 2t arrives, agents re-optimize for period 2t and 3t , based on new
information about period 2t . Agents have perfect foresight for the next 2 periods, and
adaptively generate expectations for the future thereafter.
B. Production
There are eight factors of production and three types of financial assets:
1-5. Capital types
6. Urban labor
7. Rural labor
8. Land
9. Domestic currency
10. Bank deposits
11. Foreign currency.
15 This methodology using discrete time has been in use since early work on inventory policy by Arrow, Harris and
Marschak (1951) was generalized by Bellman (1953) and contrasts with optimizations done in continuous time. This method allows us to calibrate our model before adjusting foreign demand.
78
The five types of capital correspond to five aggregate nonagricultural productive sectors:
A. Light Manufacturing
B. Heavy Industry
C. Electricity, Water, Sewage
D. Transport, Hotel
E. Housing, Health Services.
An input-output matrix, At, is used to determine intermediate and final production in
period t. We use a 25 x 25 Egyptian input-output matrix that is described in the next section.
Corresponding to each sector in the input-output matrix, sector-specific value added is
produced using capital and urban labor for the nonagricultural sectors, and land and rural
labor in agriculture. Hence At may have any dimension, but capital is specific to sectors
aggregated in the above way. Accordingly, capital is perfectly mobile across a given subsector,
but is immobile across other subsectors. Labor, on the other hand, may migrate from the rural
to the urban sector.16 This rural-urban migration is an important feature of the Egyptian
economy.
The specific formulation of the firm's problem is as follows. Let j
Kiy , j
Liy be the inputs of
capital and urban labor to the jth nonagricultural sector in period i. Let GiY be the outstanding
stock of government infrastructure in period i. The production of value added in sector j in
period i is then given by:
( , , ) j j
ji ji Ki Li Giva va y y Y (14)
16 We assume that the labor market is not segmented and there is no wage differential between workers in the
underground and the formal economy.
79
where public infrastructure may act as a productivity increment to private production.
Sector j pays income taxes on inputs of capital and labor, given by , Kij Lijt t respectively,
in period i. The interpretation of these taxes is that the capital tax is a tax on firm profits, while
the labor tax is a personal income tax that is withheld at source.
We suppose that each type of sector capital is produced via a sector-specific investment
technology that uses inputs of capital and labor to produce new capital. Investment is carried
out by the private sector and is entirely financed by domestic borrowing.
Let us define the following notation:
HiC = The cost of producing the quantity H of capital C of a particular type in period i.
ir = The interest rate r in period i.
KiP = The return to capital in period i.
MiP = The price of money in period i.
i = The rate of depreciation of capital.
Suppose, then, that the rental price of capital in period 1 is1P . If
1HC is the cost-
minimizing cost of producing the quantity of capital,1H , then the cost of borrowing must equal
the present value of the return on new capital. Hence:
2
11 1
2
1
(1 )
(1 )
inKi
H ii
j
j
P HC
r
(15)
80
where is the interest rate in period j, given by:
1/j Bjr P (16)
where PBj is the price of a bond in period j. The tax on capital is implicitly included in the
investment problem, as capital taxes are paid on capital as an input to production.
The decision to invest depends not only on the variables in the above equation, but also
upon the decision the firm makes as to whether it should pay taxes. This decision determines
the firm’s entry into the underground economy. We assume that the firm’s decision is based
upon a comparison of the tax rate on capital with the rate of return on new capital. If the tax
rate on capital is less than the corresponding rate of return, the firm pays the full tax. If the tax
rate is greater than the return to new capital, then the firm pays less than the full capital tax.
That is, it withdraws, at least partially, into the underground economy.
Formally, in a two period world, suppose that:
2
1
11
KK
Pt
r
(17)
such that the present value of the return on one unit of new capital is greater than the current
tax rate on capital. In this case, we assume the investor pays the full tax rate on capital inputs.
Suppose, on the other hand, that:
2
1
11
KK
Pt
r
(18)
Here the discounted rate of return is less than the tax rate on capital, and the firm will attempt
to reduce its tax payments by moving into the underground economy. The extent to which the
81
firm goes into the underground economy is determined by the gap between the tax rate and
the rate of return to investment. That is, the tax rate of the firm is 1Kt where:
21
21 1
1
11
KK
K K
K
Pt
rt t
t
(19)
Here 0 and higher values of lead to lower values of taxes actually paid. That is, the ratio
1
1
K
K
t
t reflects the share of the sector that operates in the above-ground economy. Hence
represents a firm-specific behavioral variable. An “honest” firm would set 0 , while a firm
that is prone to evasion would have a high value for . We should note that the firm does not
actually pay a “rate” lower than the statutory rate. Rather, it under-reports its income so that
the effective tax rate paid is ̅ . If a sector can avoid paying taxes, as above, by going into the
underground economy, why does it pay taxes at all? That is, why does it simply not set1 0Kt ?
In the next section we develop a simple approach that supposes that a firm’s refusal to pay
taxes reduces its ability to borrow from the commercial banking system. Thus, a firm’s desire to
invest will constrain its evasion of tax payments.
Suppose that the firm carries out this strategy of reducing tax payments, based on its
honesty coefficient. We suppose that the firm’s goal is to maximize the real value of capital
stock in its final period, T. Accordingly, the firm wishes to maximize ̅ where:
82
[( ) ∑ ( )
]
(20)
and is the consumer price index in period T.
C. Banking
The banking sector in our model is quite simple and is meant to capture some of the key
features and problems in Egypt as well as many other developing countries. We will suppose
that there is one bank for each nonagricultural sector of the economy. There are five such
sectors, and hence five banks, corresponding to each of the sectors in the aggregate national
income accounts. Such sector specialization of the banking system reflects the reality of many
developing countries.
Each bank lends primarily to the sector with which it is associated. The banks are,
however, not fully specialized in their corresponding sector. We make the simplifying
assumption that each bank holds a fixed share of the outstanding debt of its particular sector. It
then holds additional fixed shares of the debt of each of the remaining sectors. We make this
assumption of diversification of assets in order to allow for a situation in which a firm that
evades taxes, and thereby enters the underground economy, might receive varying degrees of
credit rationing from the different banks to which it applies for loans.
We choose a simple approach to determine the degree of credit rationing that firms
face. Our premise is that banks have no direct way of knowing whether specific firms operate in
83
the underground economy. We assume that banks only care about the estimated amount of
capital the firm may have. If the firm defaults on its loan, then this represents the best estimate
of the amount that the bank could seize. The bank would, presumably, be willing to lend an
amount equal to at least the firm’s estimated capital. If the firm requests a loan larger than its
estimated capital, the bank may choose to grant the full loan, or it may choose to restrict the
loan amount. This restriction would depend, in turn, upon the bank’s degree of risk aversion.
How can the bank estimate the value of the firm’s capital if this information is not
directly revealed by the firm? We assume the borrower is required to show the bank his tax
returns in order to obtain a loan. There is a single, flat corporate tax rate that the borrowing
firm faces. Hence, suppose that 1KT represents taxes actually paid by the borrower in period 1.
This is known to the bank, as the potential borrower is required to present his tax returns. Thus,
if the borrower is in full compliance with his tax obligation, and hence carried out no
underground activity, the value of his capital,1K̂ , would be given by:
1
1
1
ˆ K
K
TK
t (21)
Accordingly, the bank would be willing to lend at least 1K̂ to the borrower, as this would
represent a minimum estimate of the value of the firm’s capital, which could be seized in the
84
event of a default.17 Suppose, however, that the amount the firm wishes to borrow, , as in
equation (15), such that:
1 1ˆ
HC K (22)
In this case the bank lends an amount1L , where
1 1HL C , as the bank would not be able to
seize the full value of the loan in the case of a default. The situation we have described would,
in the case of perfect certainty, prompt credit rationing when the estimated value of the firm’s
capital is less than its loan request. If the firm’s capital is greater than its loan request, there
would be no credit rationing.
In a more realistic case of uncertainty about both the true value of the firm, as well as
about the bank’s own ability to seize the firm, one might expect the lending process to be
somewhat different. Accordingly, we will suppose that a simple functional form determines
bank lending as a function of the amount requested as well as the estimated value of the firm’s
capital. We define the amount the bank lends,1L , as:
1
111 1 1
1 1 1
1
ˆ
ˆ
ˆ ˆ1
HH H
H
H
K
KCL C C
K C K
C
(23)
17 We have not explicitly incorporated bankruptcies and defaults in this model, for the sake of simplicity. However,
bankruptcies and corresponding bank contractions can be introduced as in Ball and Feltenstein (2001) and Blejer, Feldman, and Feltenstein (2002).
1HC
85
Here represents a measure of risk aversion by the bank. If 0 , there are no credit
restrictions, and the bank ignores estimates of the borrower’s estimated net worth. As rises,
the bank increasingly restricts lending if the term in brackets is less than 1. If the firm pays no
taxes, hence operating entirely in the underground economy,1ˆ 0K and hence
1 0L , that is
there is no lending. If 1
1
ˆ
H
K
C increases, as would be the case if the value of the firm increases
relative to its borrowing request, then 1 1HL C , that is, the bank lends the full value of the
request.
Thus, if a firm operates entirely in the underground economy it will not be able to
borrow to finance investment. If banks are highly risk averse, they will never lend more than a
firm’s estimated net worth, which is based on its tax return. This tax return, therefore,
represents all the information the bank needs in order to determine its response to a request
for a loan.
D. Consumption
There are two types of consumers, representing rural and urban labor.18 We further
divide the two consumer categories into quintiles. We suppose that each of the now ten
representative consumers has differing Cobb-Douglas demands. The consumers also differ in
18 We use two consumer categories in order to correspond to available country data classifications, as described in
the next section.
86
their initial allocations of factors and financial assets. Cobb-Douglas parameters, as well as
factor and asset allocations, are parameterized according to the SAM. There is no
differentiation between quintiles of either urban or rural labor in the production function; so,
to a firm, there are only two sources of labor, rather than the ten we have simulated for
consumer income analysis.
The consumers maximize intertemporal utility functions, which have as arguments the
levels of consumption and leisure in each of the two periods. We permit rural-urban migration,
which depends upon the relative rural and urban wage rate. The consumers maximize these
utility functions subject to intertemporal budget constraints. The consumer saves by holding
money, domestic bank deposits, and foreign currency. He requires money for transactions
purposes, but his demand for money is sensitive to changes in the inflation rate. The consumer
pays taxes on his consumption, and does not have any direct contact with the underground
economy. That is, he pays the full nominal rates under all circumstances19.
E. Government
The government collects personal income, corporate profit, and value-added taxes, as
well as import duties. It pays for the production of public goods, as well as for subsidies. In
19 Refer to Dabla-Norris and Feltenstein for the specific form or the consumer’s problem.
87
addition, the government must cover both domestic and foreign interest obligations on public
debt. The central government deficit is defined by:20
1 1 1 1 0 1 1 0 1F FD G S rB r e B T (24)
where,
D1 = government deficit in period 1,
G1 = spending on goods and services in period 1,
S1 = subsidies given in period 1,
r1 B0 = interest obligations on the stock of domestic debt,
rF1 e1BF0 = interest obligations on the stock of foreign debt,
T1 = tax revenue in period 1.
Note that the tax revenue is partially determined by the degree to which firms enter or exit the
formal sector.
The resulting deficit is financed by a combination of monetary expansion, as well as
domestic and foreign borrowing.
2 2 2 2 1 0 2 2 1 0 2( ) ( )BG F F FD G S r y B e r C B T (25)
where,
D2 = government deficit in period 2,
20 As before, 1 denotes period i and 2 denotes period i+1.
88
G2 = spending on goods and services in period 2,
S2 = subsidies given in period 1,
2 1 0( )BGr y B = interest on initial domestic debt stock plus period 1 borrowing,
2 2 1 0( )F F Fe r C B = interest on initial foreign debt stock plus period 1 foreign borrowing,
T2 = tax revenue in period 2.
The government finances its budget deficit by a combination of monetization, domestic
borrowing, and foreign borrowing. We assume that foreign borrowing in period i, CFi, is
exogenously determined by the lender. The government then determines the face value of its
bond sales in period i, ΔyBGi, and finances the remainder of the budget deficit by monetization.
Hence:
i Bi BGi Mi Mi i FiD P y P y eC (26)
F. The Foreign Sector
The foreign sector is represented by a simple export equation in which aggregate
demand for exports is determined by domestic and foreign price indices, as well as world
income. The specific form of the export equation is:
1
0 1 2n wi
i Fi
X ye
(27)
where,
89
ΔXno = change in the dollar value of exports in period i,
πi = inflation in the domestic price index,
ie = percentage change in the exchange rate,
πFi = foreign rate of inflation,
Δywi = percentage change in world income, denominated in dollars,
σ1 = real foreign income elasticity,
σ2 = real exchange rate elasticity of export demand.
The combination of the export equation and domestic supply responses determines
aggregate exports. Demand for imports is endogenous and is derived from the domestic
consumers' maximization problems. Foreign lending is assumed to be exogenous. Thus, gross
capital inflows are exogenous, but the overall change in reserves is endogenous. Finally, we will
suppose that the exchange rate is fixed.
V. DATA SOURCES
In order to simulate our model we have used a variety of data sources and parameter
estimates for Egypt. We will first describe the sources of data regarding production,
consumption, effective tax rates, initial stocks, and the estimation of behavioral equations
before we turn to the implementation of the model.
90
A. Production
The input-output structure of intermediate and final production is derived from the
Social Accounting Matrix (SAM) from 2008. This SAM was provided by Cairo University and is
the product of the Egyptian Central Agency for Public Mobilization and Statistics (CAPMAS)21.
Our input-output matrix is aggregated to 25 x 25 with an additional row and column for foreign
trade.
Real value added per unit output for each of the 25 domestic sectors is derived from
the corresponding shares of wages and gross operating surpluses in each sector's value added.
The government is assumed to have a Cobb-Douglas production function whose coefficients are
those of the aggregate economy. Finally, in the absence of direct estimation of investment
functions, we have taken the functions to be the same for each type of capital. The coefficients
of these functions are taken to be those of the value added function for the construction
industry. Even though the functions are the same, it does not imply that the levels of
investment in the different capital types will be identical. This is because investment depends
on the interest rate and the rate of return to capital, which may differ across capital types.
21 See Appendix E for information regarding the Egyptian SAM.
91
B. Effective Tax Rates
We use direct sources from Egypt for our tax structures. Sources include: Law of
customs Duties No66 of year 1963, Social Insurance Law no 79 for 1975, Law of Stamp Duties
No 111 of year 1980, Law of Sales Taxes No 11 of year 1991 and No 2 of year 1997, Law of
Income Taxes No 91 of year 2005, Treasury Bill Law No 114 for 2008, Decree of Ministry of
Finance no 157 for 2010, Social Insurance Law No 135 of year 2010, Sales Tax Law 49 for 2011.22
C. Consumption
There are two domestic consumer categories in our model: urban and rural. Each of
these consumer groups is divided into quintiles for consumer income analysis. We take
consumption weights on each of the 26 input-output goods (including foreign trade) as the
expenditure shares in the input-output matrix. We assume that there is a single foreign
consumer, representing the rest of the world. These consumers’ demand weights are given by
export expenditure shares.
D. Initial stocks
22 Our late colleague Manal Metwaly of Cairo University was instrumental in helping us overcome the language
barrier to collect this data.
92
In our simulations, all initial allocations of factors and financial assets are taken to be
stocks at the end of 2008. Stocks of urban and rural labor are obtained by applying the shares
of income going to urban and rural labor by quintile (derived from the Egypt SAM) to 2008
Egyptian GDP. Money stocks are taken as M2 for 2008, while initial holdings of interest bearing
assets are taken as total government domestic debt, all derived from Central Bank of Egypt’s
Monthly Statistical Bulletin. Foreign assets are taken from the same source. The stock of land is
taken to be the real value added to agriculture and is taken from the Egypt SAM as the gross
operating surplus of the agricultural sector in 2008. Finally, capital stocks are determined as
the gross operating surpluses of the corresponding aggregate sectors in the Egyptian SAM.
E. Estimation of Behavioral Equations
Here are the estimations of the behavioral parameters of money demand and export
supply for the Egyptian economy. An export demand equation is not estimated here due to
Egypt’s small share in the world trade. We use estimation techniques that have been used in
the past for other developing countries.
93
The demand for money is taken from El-Shazly, 2009. The specific form we use is the
long run demand for real broad money
( )
(28)
The terms on the right hand side of the equation are:
= real income,
= domestic interest rate,
= foreign interest rate,
= real exchange rate,
= inflation rate.
All of these are generated within our simulations, with the foreign interest rate being
exogenous.
For the export supply equation, we use Ikram, 2006, in which an Egyptian export price
elasticity of -1.03 is given, while the world income demand elasticity is approximately 1.0. We
assume that total world expenditure on Egyptian exports is divided into demand for the input-
output goods according to the shares of exports from each sector in the SAM.
94
VI. SIMULATIONS
There are three simulations that we refer to as: baseline, revolution, and revolution with
tax cut. The model is calibrated to the Egyptian economic data for the years 2008—the year of
the SAM, to 2010—the year for the national accounts, and using some new data from 2011 and
onward regarding tax changes. The baseline model includes no exogenous changes to any
variables or implementation of alternative policies. The revolution simulation includes a 70%
decrease in the foreign demand for those industries that were hit hardest by the revolution:
tourism, transportation, hotels and restaurants, as well as trade and finance. This adjustment is
conducted for a single time period to isolate the effect of the foreign trade aspect of the
revolution in that year and on the long term growth path. The final simulation, revolution with
tax cut, is identical to the revolution treatment except the corporate income tax is reduced by
25% for the duration of the model to study the effects of the revolution on the economy with
an alternative fiscal policy that encourages higher tax compliance.
VII. RESULTS
In analyzing these results we look at the macroeconomic effects on the economy and
the government’s budget, the effects on consumers’ utility and income, and the effects on the
sectors’ participation in the formal economy.
95
A. Gross Domestic Product
Regarding the growth rate, in Figure 1 we find that our model, absent the revolution,
predicts a return to the robust growth rates seen in the 2005-2010 time period when our SAM
was built. However, Figures 1 and 2 reveal that with the revolution the real GDP growth rate is
4.3% less than the baseline in the absence of the lower tax on capital. With the lower
corporate income tax rate, we see a 3% net decrease from the higher growth path. The long
run effect on GDP of the one-year change in foreign demand is that, five years after the
revolution, real GDP is diminished by 3.25%. The effect on GDP is lessened by a more robust
growth path in the presence of a cut in corporate income taxes.
FIGURE 1 – Real GDP Growth Rates
2 3 4 5 6 7 8
Baseline 4.15% 2.38% 10.42% 8.84% 8.09% 9.07% 5.21%
Revolution 4.15% -1.88% 10.56% 8.62% 8.40% 9.40% 5.67%
Revolution and Tax Cut 4.01% -0.61% 10.05% 9.76% 8.00% 9.49% 5.61%
-4.00%
-2.00%
0.00%
2.00%
4.00%
6.00%
8.00%
10.00%
12.00%
96
FIGURE 2 – Normalized Real GDP Growth Rate
B. Inflation
Regarding inflation, our simulation results again show a return to the high inflation
found in 2008-10 during the time our data was extracted and inflation averaged 13.6%. Our
baseline inflation rate averages 10.9% and ranges from 5% to 17.4%. Inflation in the course of
the revolution simulation is lower, with an average of 10.6%, while the final treatment,
including the tax cut, shows an average inflation rate of 11.5%. Figure 3 reveals the increasing
inflation and the effect of the revolution and tax cut on inflation over the duration of the
simulation.
92
93
94
95
96
97
98
99
100
101
102
1 2 3 4 5 6 7 8
No
rmal
ize
d G
DP
Baseline
Revolution
Revolution andTax Cut
97
FIGURE 3 – Annual Inflation Rates
C. Budget
Figure 4 shows that the revolution has a negligible effect on tax collections as a percent
of GDP, although they are slightly lower as a result of decreased tax revenues from the effected
sectors of the economy, but this effect is offset by the smaller economy in that simulation. As
we would expect, the lower tax rate on corporations does diminish the tax revenue as a percent
of the overall economy. However, the decrease in revenue is not the absolute amount that a
static analysis would reveal, as additional firms enter the formal economy and begin paying
taxes on their capital in order to access capital markets. Figure 4 also shows that the model
predicts a slight increase in the tax-to-GDP ratio from approximately 19% to a bit less than 21%.
2 3 4 5 6 7 8
Baseline 4.95% 5.10% 9.79% 11.68% 12.43% 15.75% 17.39%
Revolution 4.95% 5.74% 9.71% 10.85% 12.50% 16.71% 14.90%
Revolution and Tax Cut 5.32% 6.67% 9.87% 12.71% 12.58% 18.45% 15.18%
0%
2%
4%
6%
8%
10%
12%
14%
16%
18%
20%
98
FIGURE 4 – Tax Revenue to GDP Ratio
Regarding deficits, we see the government running increasingly large deficits into the
future (Figure 5). The problem of the deficit is exacerbated by the revolution and further
compounded by the tax cut.
1 2 3 4 5 6 7 8
Baseline 18.9% 19.8% 19.8% 19.9% 20.3% 20.4% 20.8% 20.9%
Revolution 18.9% 19.8% 19.9% 20.0% 20.3% 20.5% 20.8% 20.9%
Revolution and Tax Cut 17.6% 18.5% 18.8% 18.8% 19.1% 19.1% 19.4% 19.5%
17.0%
18.0%
19.0%
20.0%
21.0%
99
FIGURE 5 – Deficit as a Percent of GDP
D. Consumers
In Table 2, we see the real incomes of the ten consumer groups in period 8. Consumers
1-5 are urban, consumers 6-10 are rural, and the poorest quintiles are 1 and 6. The first thing
to note is that urban consumers are considerably wealthier than their rural counterparts. In fact,
all but the poorest urban quintile are richer than the richest rural quintile. We also see that the
simulated revolution affects urban consumers especially hard, while rural consumers are
1 2 3 4 5 6 7 8
Baseline -5.50% -4.60% -15.40% -14.20% -21.40% -21.70% -24.40% -27.30%
Revolution -5.50% -4.60% -15.90% -15.10% -22.90% -23.50% -25.90% -28.50%
Revolution and Tax Cut -6.90% -6.10% -17.00% -16.60% -23.90% -25.00% -27.90% -30.70%
-35%
-30%
-25%
-20%
-15%
-10%
-5%
0%
100
actually helped, though the high percentages are an artifact of their low base. This result occurs
because the diminished foreign demand almost exclusively affects urban labor, provided by
urban consumers. The diminished demand for goods produced using urban labor results in a
diminished relative urban wage. We also see that the increased growth provided by the lower
tax rate on capital brings urban consumers back to the baseline while increasing the gains made
by rural consumers.
TABLE 22 – Real income in period 8 by consumer
Baseline Revolution Revolution with Tax Cut
Income Income % Change Income % Change
Urban
Consumer 1 14.6 13.8 -5.33% 14.6 -0.02%
Consumer 2 21.3 20.1 -5.36% 21.3 -0.04%
Consumer 3 26.5 25.2 -5.23% 26.6 0.09%
Consumer 4 35.2 33.5 -4.92% 35.4 0.40%
Consumer 5 76.6 74.3 -3.03% 78.3 2.26%
Rural
Consumer 6 3.5 3.9 13.34% 4.1 18.38%
Consumer 7 4.9 5.5 12.30% 5.8 17.37%
Consumer 8 6.1 6.8 11.96% 7.1 17.01%
Consumer 9 7.6 8.5 11.63% 8.9 16.69%
Consumer 10 19.6 21.3 8.36% 22.2 13.00%
In Table 3, we see the utility values in period 8 relative to the baseline, which is
normalized to 100. Again, we see in the revolution simulation that rural consumers gain from
the increased relative wage rate and urban consumers suffer. However, with the inclusion of
the tax cuts, losses for urban consumers are largely diminished or reversed, while rural
consumers have increased gains.
101
TABLE 23 – Utility in period 8 by consumer
Baseline Revolution Revolution with Tax Cut
Consumer Utility Utility % Change Utility % Change
Urban Consumer 1 100.0 88.8 -11.2% 95.2 -4.8%
Consumer 2 100.0 92.3 -7.7% 98.5 -1.5%
Consumer 3 100.0 93.8 -6.2% 99.9 -0.1%
Consumer 4 100.0 95.0 -5.0% 101.0 1.0%
Consumer 5 100.0 98.9 -1.1% 104.7 4.7%
Rural Consumer 6 100.0 121.2 21.2% 127.3 27.3%
Consumer 7 100.0 120.8 20.8% 126.7 26.7%
Consumer 8 100.0 121.6 21.6% 127.3 27.3%
Consumer 9 100.0 121.9 21.9% 127.6 27.6%
Consumer 10 100.0 116.3 16.3% 120.9 20.9%
E. Formal Sector
Our final variable of interest is the level to which sectors are entering or leaving the
formal sector of the economy. Figures 6 and 7 show the percentage participation of each
capital sector in the formal economy relative to the baseline for the two alternative treatments.
In Figure 6, we see that Heavy Industry and Electricity, Water, and Sewage decrease their levels
of compliance and thus operate in the informal sector at a greater rate as a result of the
revolution. The other industries are not affected.
102
FIGURE 6 – Tax Compliance Relative to the Baseline: Revolution
Figure 7 shows the effect on participation in the formal economy of the capital sectors
relative to the baseline in the final treatment. Here we see that, despite the revolution, the
decreased tax on capital has an unambiguously positive impact on participation in the formal
economy, as all capital sectors operate in the formal sector at an equal or greater rate. This
affirms our theory’s prediction, given that the lower tax rate increases the return to capital, and
thus makes borrowing capital and paying taxes to verify assets more attractive. Also of note is
that transportation and light manufacturing are not affected in either simulation.
2 4 6 8
Light Manufacturing 0% 0% 0% 0%
Heavy Industry 0% -6% -15% 0%
Electricity, Water, Sewage 0% -10% -19% -32%
Transport 0% 0% 0% 0%
Hotels, Housing, HealthServices
0% 0% 0% 0%
-35%
-30%
-25%
-20%
-15%
-10%
-5%
0%
103
FIGURE 7 – Tax Compliance Relative to the Baseline: Revolution with Tax Cut
VIII. CONCLUSIONS
From the treatments, we see the long run negative effects and consequences of the
revolution on the Egyptian economy. We also see that some of the negative effects can be
mitigated by implementing a lower tax on capital and drawing more economic activity into the
formal economy. We expect the economic impact of the revolution to be particularly hard on
urban workers, as their industries were hardest hit by diminished foreign demand for those
goods that they produced. We also expect macroeconomic trends to worsen as the duration of
2 4 6 8
Light Manufacturing 0% 0% 0% 0%
Heavy Industry 37% 56% 1% 0%
Electricity, Water, Sewage 16% 2% 9% 3%
Transport 0% 0% 0% 0%
Hotels, Housing, HealthServices
5% 0% 0% 0%
0%
10%
20%
30%
40%
50%
60%
104
the revolution lengthens beyond our single time period simulation. One important policy
implication of this research is that a significant portion of the economic damage done by the
revolution could be mitigated with increased investment spurred by a lower corporate tax rate.
However, this would exacerbate some budgetary pressures, though not by the amount
indicated by a static analysis, as tax compliance and growth would both be greater.
Extensions of this research may include alternative monetary policies or more
comprehensive fiscal policies. Future researcher may also consider alternative uses of the large
foreign aid packages being pledged to Egypt. The most obvious extension would expand the
shock of the revolution to include multiple time periods representing specific years and
calibrated to new data on the Egyptian economy. This analysis may include a comparison of
optimistic and pessimistic assumptions about the return of foreign demand.
105
CHAPTER FOUR: SUMMARY AND CONCLUSIONS
The objective of this research was to investigate the effects of alternative fiscal policies
on growth, welfare, and budgets in two interesting and very different cases and then contrast
the variations in methodology and findings of the two applications. Regarding the findings, in
Georgia we found that a revenue neutral tax reform in direction of greater efficiency negligibly
improved gross state product and had a negligibly regressive impact on households, while a
more drastic tax reform in the same direction negligibly decreased state product and had a
significantly regressive impact on households. We found that the effects on industries of
implementing a uniform tax harms those who had paid a below average rate and helps those
who had paid an above average rate, but that in the more drastic tax reform the magnitude of
the effects is greater with a net positive effect on output greater than we found in the milder
reform.
In Egypt, we found that the effects of a brief decrease in the foreign demand for tourism
related industries has long lasting consequences. The negative effects of the foreign demand
decrease can be better absorbed by an economy with a lower corporate income tax and the
resulting larger formal sector. We also find that the effects of the revolution hit the relatively
richer urban households significantly harder than the poorer rural households.
Contrasting the findings on growth in the two applications, we can see that the lower
taxes in have an ambiguous effect: Egyptian GDP growth rates are improved in just as many
years as they are decreased, while in Georgia the growth rates are positive in the modest
106
reform and negative in the more comprehensive reform. We also see that the tax changes in
Egypt are significantly progressive while the tax changes in Georgia are mildly regressive.
Regarding the designs, we show the flexibility of the methodology for adapting to
sensitive analysis of tax reform in a stable economy or for comparing alternative futures in light
of a regime change. We approached this by presenting various decisions modelers face in the
including the level of policy control, trading circumstances, economic stability, etc. We present
the basic alternatives available, which we chose for our purposes, why, and how.
This research extends the existing literature in a number of ways. The analysis of
Georgia models the joint effects of a tax reform package that includes adjustments to the
personal and corporate income tax, the sales tax, and taxes on intermediate goods in a regional
CGE model. No previous study has examined this combination of fiscal changes. The model also
utilizes a unique aggregation of industries to isolate industry specific tax changes and the
consequent adjustments to output by industry. We also look at the effects of an incremental
step towards an income tax elimination compared to the complete elimination. We simulate
the short and long run effects of each policy to analyze the adjustment paths relative to the
mobility and flexibility of factors. Finally, we extend the basic framework of the model
originally developed by Lofgren, et al. and present the calibration techniques and code for
future modelers.
In Egypt, we apply an existing CGE model involving tax evasion to simulate the future
growth path for Egypt after a decrease in foreign demand using the newest SAM data available.
No previous study has been published studying the post revolution Egyptian economy in this
manner. We look at the immediate and long run effects of the fall in foreign demand on
107
macroeconomic variables and on the welfare of urban and rural quintiles. We study the effects
of the resulting also simulate the regime change in alternative fiscal policy environments to
compare macroeconomic indicators.
Finally, this comparison of computable general equilibrium models in very different
economies may serve as a useful guide to future modelers interested in a fiscal policy
application. In the literature, this methodology has been used primarily for trade analysis and
environmental applications such as carbon emissions. However, this methodology is equally
useful in fiscal policy analysis, evaluation, and comparison. With what we hope is a useful
contrast of CGE application in very different economies, some future modeler may encroach on
the private sector and think-tank dominance of CGE fiscal policy analysis.
108
APPENDIX
APPENDIX A – INCOME TAX CALIBRATION CHECK
In Appendix Table A1, note that the difference between the ratio of the reformed tax
rate to the statutory tax rate and the ratio of the baseline tax rate to the simulated tax rate is
near zero. This is a check to confirm that we are accurately simulating the tax reform.
TABLE A1- Check for accuracy of income tax calibration
HHD1 HHD2 HHD3 HHD4 HHD5 HHD6 HHD7 HHD8 HHD9
Baseline ty 0.50% 0.20% 1.20% 2.00% 2.30% 3.20% 3.50% 3.70% 3.70%
Simulation ty 0.30% 0.20% 1.10% 1.90% 2.00% 2.40% 2.60% 2.70% 2.50%
Statutory (Single) 1.30% 2.70% 4.00% 4.60% 5.00% 5.30% 5.50% 5.70% 5.90%
Statutory (M2K) 0.00% 0.50% 2.30% 3.40% 4.20% 4.80% 5.10% 5.40% 5.80%
Average (Statutory) 0.70% 1.60% 3.20% 4.00% 4.60% 5.10% 5.30% 5.50% 5.80%
Reform (Single) 0.80% 2.40% 4.00% 4.00% 4.00% 4.00% 4.00% 4.00% 4.00%
Reform (M2K) 0.00% 0.00% 2.10% 3.80% 3.70% 3.80% 3.80% 3.90% 4.00%
Average (Reform) 0.40% 1.20% 3.10% 3.90% 3.80% 3.90% 3.90% 3.90% 4.00%
Ratio (Reform/Stat) 0.624 0.735 0.962 0.978 0.837 0.769 0.74 0.714 0.685
Baseline income tax rate 5.20% 1.30% 3.40% 4.00% 4.00% 4.30% 4.80% 5.00% 7.20%
Simulation income tax rate 3.30% 1.00% 3.30% 4.00% 3.30% 3.30% 3.50% 3.50% 5.00%
Ratio (Baseline/Sim) 0.629 0.74 0.967 0.985 0.839 0.77 0.74 0.713 0.684
Difference 0.005 0.005 0.005 0.006 0.002 0.001 0 -0.001 0
109
APPENDIX B - SPECIAL COUNCIL SIMULATION GAMS CODE
*Income Tax (SG/TOTAL) ty('SGOVNE','HHD1')= .00317; ty('SGOVNE','HHD2')= .00154; ty('SGOVNE','HHD3')= .01135; ty('SGOVNE','HHD4')= .01938; ty('SGOVNE','HHD5')= .01966; ty('SGOVNE','HHD6')= .02427; ty('SGOVNE','HHD7')= .02596; ty('SGOVNE','HHD8')= .02670; ty('SGOVNE','HHD9')= .02538; *Sales Tax (GA Sales Tax Rate expanded to untaxed sectors until rate is 4%) tq("AFFH-C") = 0.0; tq("MIN-C") = 0.0; tq("MANUF-C") = 0.0; tq("MEDIMAN-C") = 0.0; tq("UTIL-C") = 0.0; tq("FOOD-C") = 0.034293; tq("INFOSER-C") = 0.010062; tq("HEEDSER-C") = 0.0; tq("OTHSER-C") = 0.021532; tq("CONST-C") = 0.0; tq("WHOLTRAD-C") = 0.0; tq("RETTRAD-C") = 0.0; tq("TRAN-C") = 0.0; tq("MOVIE-C") = 0.0; tq("GOVSPC-C") = 0.0; *Intermediate Inputs Tax tq_inp("UTIL-A","AFFH-C") = -.04; tq_inp("UTIL-A","MIN-C") = -.04; tq_inp("UTIL-A","MANUF-C") = -.04; tq_inp(A ,"HEEDSER-C") = .04; tq_inp("MOVIE-A" ,"MOVIE-C") = .04; *Corporate Income Tax taxc("SGOVNE","CAP")= .004427*.66;
APPENDIX C - INCOME TAX ELIMINATION GAMS CODE
*Income Tax (SG/TOTAL)
110
ty('SGOVNE',H)= 0 *Sales Tax (GA Sales Tax Rate increased to a uniform 9% rate) tq("AFFH-C") = 0.05; tq("MIN-C") = 0.05; tq("MANUF-C") = 0.05; tq("MEDIMAN-C") = 0.05; tq("UTIL-C") = 0.05; tq("FOOD-C") = 0.084293; tq("INFOSER-C") = 0.060062; tq("HEEDSER-C") = 0.05; tq("OTHSER-C") = 0.071532; tq("CONST-C") = 0.05; tq("WHOLTRAD-C") = 0.05; tq("RETTRAD-C") = 0.05; tq("TRAN-C") = 0.05; tq("MOVIE-C") = 0.05; tq("GOVSPC-C") = 0.05; *Intermediate Inputs Tax (Tax treatment of intermediate inputs is identical to Special Council *simulation with the exception of the rate increase) tq_inp("UTIL-A","AFFH-C") = -.04; tq_inp("UTIL-A","MIN-C") = -.04; tq_inp("UTIL-A","MANUF-C") = -.04; tq_inp(A ,"HEEDSER-C") = .09; tq_inp("MOVIE-A" ,"MOVIE-C") = .09; *Corporate Income Tax taxc("SGOVNE","CAP")= .004427*0;
111
APPENDIX D – GEORGIA SAM
The Georgia 2010 Social Accounting Matrix is 53 rows x 53 columns, and thus too large
for inclusion here, but it is available upon request. The original unaggregated IMPLAN data is
available for purchase at https://implan.com/.
APPENDIX E – EGYPTIAN SAM
The Egypt 2008 Social Accounting Matrix is 72 rows x 72 columns, and thus too large for
inclusion here, but it is available upon request. For I-O and supply-use matrices from CAPMAS
for Nov. 2010 and Sept. 2008, visit http://www.capmas.gov.eg/pages_ar.aspx?pageid=1475.
112
BIBLIOGRAPHY
Abdel-Kader, Khaled. Private Sector Access to Credit in Egypt: Evidence Frome Survey Data: Egyptian Center for Economic Studies, 2006.
Achy, Lahcen. "The Uncertain Future of Egypt’s Economic Reforms." Ahram Online 22, (2010).
al-Dīn, H.K., S. Fawzy, A. Refaat, Iran Economic Research Forum for the Arab Countries, and
Turkey. Investment Incentives, Marginal Effective Tax Rates and the Cost of Capital in Egypt: Economic Research Forum, 2000.
Alissa, Sufyan. The Political Economy of Reform in Egypt: Understanding the Role of Institutions:
Carnegie Endowment for International Peace, 2007.
Arrow, Kenneth J, Theodore Harris, and Jacob Marschak. "Optimal Inventory Policy."
Econometrica: Journal of the Econometric Society, (1951): 250-272.
Ball, Sheryl Beth, and Andrew Feltenstein. Basic Macroeconomic Options for Bangladesh : A
Numerical Analysis, 1998.
Barbé, André J, and George R Zodrow. "The Relative Efficiency Costs of State Gross Receipts
Taxes and Retail Sales Taxes." In 106th Annual Conference on Taxation Tampa FL: National Tax Association, 2013.
Bellman, Richard. "Some Functional Equations in the Theory of Dynamic Programming."
Proceedings of the National Academy of Sciences of the United States of America 39, no. 10 (1953): 1077.
Berck, Peter, Elise Golan, and Bruce Smith*. "State Tax Policy, Labor, and Tax Revenue Feedback
Effects." Industrial Relations: A Journal of Economy and Society 36, no. 4 (1997): 399-418.
Blejer, Mario I, Ernesto V Feldman, and Andrew Feltenstein. "Exogenous Shocks, Contagion, and
Bank Soundness: A Macroeconomic Framework." Journal of International Money and Finance 21, no. 1 (2002): 33-52.
113
CAPMAS. "The Central Agency for Public Mobilisation and Statistics - Egypt." Tourism Statistics, (1973).
Council, World Travel and Tourism. "Economic Impact Report 2010." 2010.
Dabla-Norris, E. R. A., and Andrew Feltenstein. "The Underground Economy and Its
Macroeconomic Consequences." The Journal of Policy Reform 8, no. 2 (2005): 153-174.
Dakhlia, Sami, and Robert P Strauss. "Taxation of Producer and Consumer Purchases: What Can
Cge Tell Us About the Real World of Sales and Use Taxes?", (2003).
De Soto, Hernando. "Egypt's Economic Apartheid." Wall Street Journal 3, (2011).
Dewatripont, Mathias, and Gilles Michel. "On Closure Rules, Homogeneity and Dynamics in
Applied General Equilibrium Models." Journal of Development Economics 26, no. 1 (1987): 65-76.
Dobronogov, Anton, and Farrukh Iqbal. Economic Growth in Egypt: Constraints and
Determinants: World Bank Washington, 2005.
Dwyer, Larry, Peter Forsyth, and Ray Spurr. "Assessing the Economic Impacts of Events: A
Computable General Equilibrium Approach." Journal of Travel Research 45, no. 1 (2006): 59-66.
Edmiston, Kelly D. "Strategic Apportionment of the State Corporate Income Tax: An Applied
General Equilibrium Analysis." National Tax Journal, (2002): 239-262.
El-Shazly, Alaa. "Investment under Tax Policy Uncertainty: A Neoclassical Approach." Public
Finance Review 37, no. 6 (2009): 732-749.
Feltenstein, Andrew, and Anwar Shah. "General Equilibrium Effects of Investment Incentives in
Mexico." Journal of Development Economics 46, no. 2 (1995): 253-269.
FEMISE. "Egypt Country Profile: The Road Ahead for Egypt." In Cairo: The Economic Research
Forum for Arab Countries, Iran and Turkey (ERF), 2004.
Gallaway, Michael P, Christine A McDaniel, and Sandra A Rivera. "Short-Run and Long-Run
Industry-Level Estimates of Us Armington Elasticities." The North American Journal of Economics and Finance 14, no. 1 (2003): 49-68.
114
Giesecke, James, Peter B. Dixon, and Maureen T. Rimmer. "Regional Macroeconomic Outcomes
under Alternative Arrangements for the Financing of Public Infrastructure*." Papers in Regional Science 87, no. 1 (2008): 3-31.
Gordon, Roger, and Wei Li. "Tax Structures in Developing Countries: Many Puzzles and a
Possible Explanation." Journal of Public Economics 93, no. 7 (2009): 855-866.
Gupta, Dipak K., M. C. Madhavan, and Andrew Blee. "Democracy, Economic Growth and
Political Instability: An Integrated Perspective." The Journal of Socio-Economics 27, no. 5 (1998): 587-611.
Harris, Rebecca Lee, and Sherman Robinson. "Economy-Wide Effects of El Niño/Southern
Oscillation (Enso) in Mexico and the Role of Improved Forecasting and Technological Change." International Food Policy Research Institute, Washington DC, USA (TMD Discussion Paper 83), (2001).
Holland, David, Leroy Stodick, Stephen Devadoss, and Joydeep Ghosh. "The Economic Effect of
Mad Cow Disease and the Washington Economy." Washington State University. PowerPoint Presentation, (2004).
Horridge, Mark, John Madden, and Glyn Wittwer. "The Impact of the 2002–2003 Drought on
Australia." Journal of Policy Modeling 27, no. 3 (2005): 285-308.
Huang, Ming-Yuan, Janaki RR Alavalapati, and Onil Banerjee. "Economy-Wide Impacts of Forest
Bioenergy in Florida: A Computable General Equilibrium Analysis." Taiwan Journal of Forest Science 27, no. 1 (2012): 81-93.
Ikram, Khalid. The Egyptian Economy, 1952-2000: Performance Policies and Issues: Routledge,
2007.
IMF. Arab Republic of Egypt: 2010 Article Iv Consultation - Staff Report. 2010.
Julia-Wise, Roxana, Stephen C Cooke, and David Holland. "A Computable General Equilibrium
Analysis of a Property Tax Limitation Initiative in Idaho." Land Economics 78, no. 2 (2002): 207-227.
Kambou, Gerard, Shantayanan Devarajan, and Mead Over. "The Economic Impact of Aids in an
African Country: Simulations with a Computable General Equilibrium Model of Cameroon." Journal of African economies 1, no. 1 (1992): 109-130.
115
Kasimati, Evangelia. "Economic Aspects and the Summer Olympics: A Review of Related
Research." International Journal of Tourism Research 5, no. 6 (2003): 433-444.
Kasimati, Evangelia, and Peter Dawson. "Assessing the Impact of the 2004 Olympic Games on
the Greek Economy: A Small Macroeconometric Model." Economic Modelling 26, no. 1 (2009): 139-146.
Kawai, Masahiro, and Fan Zhai. "China–Japan–United States Integration Amid Global
Rebalancing: A Computable General Equilibrium Analysis." Journal of Asian Economics 20, no. 6 (2009): 688-699.
Koh, Young-Kon. "Analysis of Oklahoma's Boom and Bust Economy by Means of a Cge Model."
Oklahoma State University, 1991.
Kraybill, D, and D Pai. "Documentation for a Computable General Equilibrium Model of Ohio."
rnimeo, Department of Agricultural Economics, The Ohio State University, (1995).
Kydland, Finn E., and Edward C. Prescott. "Time to Build and Aggregate Fluctuations."
Econometrica 50, no. 6 (1982): 1345-1370.
Li, ShiNa, and Leo Jago. "Evaluating Economic Impacts of Major Sports Events – a Meta Analysis
of the Key Trends." Current Issues in Tourism 16, no. 6 (2012): 591-611.
Lofgren, Hans, Osten Chulu, Osky Sichinga, Franklin Simtowe, Hardwick Tchale, Ralph Teska,
and Peter Wobst. "External Shocks and Domestic Poverty Alleviation: Simulations with a Cge Model of Malawi." (2001).
Lofgren, Hans, Rebecca Lee Harris, and Sherman Robinson. A Standard Computable General
Equilibrium (Cge) Model in Gams. Vol. 5: Intl Food Policy Res Inst, 2002.
Lucas Jr, Robert E. "Econometric Policy Evaluation: A Critique." Carnegie-Rochester Conference
Series on Public Policy 1, no. 0 (1976): 19-46.
Madden, John R, and Matthew Crowe. "Estimating the Economic Impact of the Sydney Olympic
Games." In 38th Congress of the European Regional Science Association, August, Vienna, 1998.
116
Madden, John R. "Economic and Fiscal Impacts of Mega Sporting Events: A General Equilibrium Assessment." Public Finance & Management 6, no. 3 (2006): 346.
MENA-OECD. "Egypt National Investment Reform Agenda Workshop - Taxation Session:
Successfully Implementing a New Law." MENA-OECD Investment Programme, (2006).
Mills, Edwin S. "Misuse of Regional Economic Models, The." Cato J. 13, (1993): 29.
MOED. "Ministry of Economic Development - Egypt." Foreign Direct Investment Statistics,
(2011).
Morley, Samuel, Valeria Piñeiro, and Sherman Robinson. External Shocks and Policy Alternatives
in Small Open Economies: The Case of El Salvador. International Food Policy Research Institute (IFPRI), 2011.
Narayan, Paresh Kumar, and Biman Chand Prasad. "The Long-Run Impact of Coups on Fiji's
Economy: Evidence from a Computable General Equilibrium Model." Journal of International Development 19, no. 2 (2007): 149-160.
Pambudi, Djauhari, Nathalie McCaughey, and Russell Smyth. "Computable General Equilibrium
Estimates of the Impact of the Bali Bombing on the Indonesian Economy." Tourism Management 30, no. 2 (2009): 232-239.
Porter, Philip K., and Deborah Fletcher. "The Economic Impact of the Olympic Games: Ex Ante
Predictions and Ex Poste Reality." Journal of Sport Management 22, no. 4 (2008): 470-486.
Robinson, Sherman, and Hans Lofgren. "Macro Models and Poverty Analysis: Theoretical
Tensions and Empirical Practice." Development Policy Review 23, no. 3 (2005): 267-283.
Robinson, Sherman, Shankar Subramanian, and Jacqueline Geoghegan. "Modeling Air Pollution
Abatement in a Market-Based Incentive Framework for the Los Angeles Basin." In Economic Instruments for Air Pollution Control, 46-72: Springer, 1994.
Rockwood, Elizabeth. "The Effects on a New Marriage of Time Allocated to Economics." Journal
of Patience and Encouragement. August 24, 2013.
117
Rose, Adam, and Shu‐Yi Liao. "Modeling Regional Economic Resilience to Disasters: A Computable General Equilibrium Analysis of Water Service Disruptions*." Journal of Regional Science 45, no. 1 (2005): 75-112.
Russo, Benjamin. "An Efficiency Analysis of Proposed State and Local Sales Tax Reforms."
Southern Economic Journal, (2005): 443-462.
Sen, A. K. "Neo-Classical and Neo-Keynbsian Theories of Distribution*." Economic Record 39, no.
85 (1963): 53-64.
Seung, Chang K, and David S Kraybill. "The Effects of Infrastructure Investment: A Two-Sector
Dynamic Computable General Equilibrium Analysis for Ohio." International Regional Science Review 24, no. 2 (2001): 261-281.
Seung, Chang K, and Edward C Waters. "Evaluating Supply-Side and Demand-Side Shocks for
Fisheries: A Computable General Equilibrium (Cge) Model for Alaska." Economic Systems Research 22, no. 1 (2010): 87-109.
Seung, Chang Kyu, and David S. Kraybill. "Tax Incentives in an Economy with Public Goods."
Growth and Change 30, no. 1 (1999): 128-147.
Stodick, Leroy, David Holland, and Stephan Devadoss. "Documentation for the Idaho-
Washington Cge Model." School of Economic Sciences, Washington State University, (2004).
Tax Reform Council. "Final Report - Special Council on Tax Reform and Fairness for Georgians."
2011.
Thaiprasert, Nalitra, Dagney Faulk, and Michael J Hicks. An Examination of the Economic Impact
of Property Tax Levy Caps on the Economic Activity of New Jersey. 2011.
Thaiprasert, Nalitra, Dagney Faulk, and Michael J Hicks. "A Regional Computable General
Equilibrium Analysis of Property Tax Rate Caps and a Sales Tax Rate Increase in Indiana." Public Finance Review 41, no. 4 (2013): 446-472.
Tohamy, Sahar. "Tax Administration and Transactions Costs in Egypt." Partners for Development:
New Roles for Government and Private Sector in the Middle East and North Africa, (1999).
118
Toye, John. "Fiscal Crisis and Fiscal Reform in Developing Countries." Cambridge Journal of Economics 24, no. 1 (2000): 21-44.
Turnovsky, Stephen J, and Md A Basher. "Fiscal Policy and the Structure of Production in a Two-
Sector Developing Economy." Journal of Development Economics 88, no. 2 (2009): 205-216.
Vargas, Eliecer E, Dean Schreiner, Gelson Tembo, and David Marcouiller. "Computable General
Equilibrium Modeling for Regional Analysis." Regional Research Institute, West Virginia University Morgantown, West Virginia, 1999.
Vasche, Jon David. "Whatever Happened to Dynamic Revenue Analysis in California? ." In
Annual Revenue Estimation & Tax Research Conference, edited by Federation of Tax Administrators. Portland, Oregon, 2006.
Waters, Edward C, David W Holland, and Bruce A Weber. "Economic Impacts of a Property Tax
Limitation: A Computable General Equilibrium Analysis of Oregon's Measure 5." Land Economics, (1997): 72-89.
Watson, Philip S., Kimberly Castelin, Priscilla Salant, and J. D. Wulfhorst. "Estimating the
Impacts of a Reduction in the Foreign-Born Labor Supply on a State Economy: A Nested Cge Analysis of the Idaho Economy." Review of Regional Studies 42, no. 1 (2012): 51-74.
WEO-IMF. "World Economic Outlook." Washington, DC, 2014.
WorldBank. "International Bank for Reconstruction and Development Project Appraisal
Document for a Proposed Loan in the Amount of Us $300 Million to the Arab Republic of Egypt for an Enhancing Access to Finance for Micro and Small Enterprises Project". 2010. Social and Economic Development Group
119
VITA
Jeffrey Condon was born in New Orleans in 1980 and spent much of his childhood in
Peachtree City, Georgia. He attended college at the University of Georgia studying Finance and
Economics. Upon graduating he moved to Birmingham, Alabama to work as an auditor for
Regions Financial before returning to Atlanta as an Economist for the Bureau of Labor Statistics.
While there he attended Georgia State University for his Master of Arts in Economics. After
completing his MA he then enrolled in the PhD program at GSU and began teaching for Georgia
Military College. Mr. Condon’s fields are public finance and experimental economics. He has
researched tax reform in the state of Georgia as well as fiscal and monetary policies in Egypt
and Mauritius.